HOLOGRAPHIC FLUCTUATION MICROSCOPY APPARATUS AND METHOD FOR DETERMINING MOBILITY OF PARTICLE AND/OR CELL DISPERSIONS

Abstract

The present invention relates to an instrument and a measurement apparatus and methodology that yields a measurement and test methodology that characterizes a population of cells/particles or detects a sub-population of cells/particles based on their detected mobility in a quick and efficient manner.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS

The present invention claims priority from U.S. Provisional Patent Application No. 61/347,956, filed May 25, 2010, and U.S. Provisional Patent Application No. 61/348,072, filed May 25, 2010, the contents of both of which are herein incorporated by reference in their entirety.

BACKGROUND OF THE INVENTION

1. Field of the Invention

The present invention relates to an instrument and a measurement apparatus and methodology that yields a measurement and test method that characterizes a population of cells/particles and/or their interactions with a chemically modified (or unmodified) surface, and/or detects a sub-population of cells/particles based on their detected mobility in a quick and efficient manner.

2. Description of the Related Art

Traditional cell/particle tracking methods may utilize standard microscopy techniques such as brightfield, darkfield, and fluorescence to detect particles moving in two or three dimensions. However, at lower magnifications, there are a large number of cells/beads present in a field of view, which causes traditional frame-by-frame particle detection and tracking approaches to be slower than required. Since it is critical to minimize testing time for high-throughput diagnostic applications, for example, a new microscopy method and apparatus which can measure particle mobility more quickly, is desired.

SUMMARY OF THE INVENTION

The present invention relates to an instrument and a measurement apparatus and methodology that yields a measurement and test methodology that characterizes a population of cells/particles or detects a sub-population of cells/particles based on their detected mobility in a quick and efficient manner.

In one embodiment consistent with the present invention, an apparatus for measuring mobility of particles includes a coherent light source (e.g. laser, superluminescent diode); a collimator which collimates the light source (which may be fiber coupled); a transparent sample holder which holds a sample of particles or cells, the particles/cells which are illuminated by the laser beam; a transparent coverslip disposed on the sample holder, the coverslip which is treated with a predetermined linker molecule which attaches to an appropriate antibody/antigen on a surface of the coverslip to allow specific binding of the particles/cells; an objective lens which receives the laser beam from the transparent sample holder; and a monitor which receives the laser beam from the objective lens, the monitor which images a magnified pattern of the particles/cells, illuminated in transmission.

In one embodiment, an apparatus for measuring, testing and characterizing a population or sub-population of particles based on their detected mobility, includes: an imaging forming apparatus, including: a coherent light source which emits a light beam; and a collimator which collimates said light beam from said coherent light source; a transparent sample holder of a microscope on which a sample is disposed and which is illuminated by said collimated light beam, said sample which comprises a dispersion of particles; and means for measuring a mobility of said particles on said sample holder, in order to infer a presence or absence of interactions of said particles with said sample holder.

In a similar embodiment, a transparent, semi-transparent, or partially mirrored sample and sample chamber may be measured in reflection mode, involving the laser illumination to illuminate the sample from the collection side, and image formation occurring from the light reflected from the sample that is subsequently imaged onto the monitor.

In yet another embodiment consistent with the present invention, the transparent, semi-transparent, or partially mirrored sample holder is an automated fluidic device or microtiter plate device with data acquisition and analysis capability.

The mobility of the cells/particles may be a function of particular properties that affect the interaction between the cells/particles and the substrate (e.g., specific binding of surface antigens to surface bound antigens, hydrophobic, electrostatic etc.). The type and magnitude of the cell/particle-surface interaction will affect the extent of the thermally generated motion of cells/microscopic particles at thermal equilibrium. It will also affect the response of the cell/particle to applied physical forces. Cells/particles with negligible interactions with the surface undergo Brownian-type motion, while interactions that are sufficiently strong will tend to limit the range of motion observed on the surface (e.g., hindered Brownian motion). Similarly, cells/particles with weak interactions with the surface will have a greater response to applied physical forces, though in the presence of stronger surface interactions the response will be attenuated.

The mobility of the cells/particle dispersion is also a function of collective properties such as their effective viscosity, visco-elasticity etc., in the medium measured at thermal equilibrium. Cells/particles in a more viscous medium for example, will also have their thermally (or physically) driven range of motion attenuated compared to identical cells/particles in a less viscous medium.

The present invention characterizes dispersions of cells/particles based on their detected diffusional properties (e.g., effective diffusion coefficient(s), effective viscosity) measured at thermal equilibrium or in the presence of applied perturbing forces.

As an alternative to traditional cell/particle tracking methods, the holographic fluctuation microscopy apparatus and method (technique) described herein is readily applicable to low-magnification measurements (allowing increased throughput) and has less stringent focusing requirements. An additional advantage of this apparatus and technique is that due to the possibility of imaging the sample with coherent illumination, significantly out-of-focus images of the samples' diffraction patterns may be imaged. The diffraction patterns may then be mathematically transformed and numerically propagated to calculate an extremum for a focus measure (see W. Li et al., J. Opt. Soc. Am. A, Vol. 24, No. 10, 3054-62, 2007, for example). The numerically propagated distance is related to the distance from focus, which is determined by the use of a calibration curve that relates the location of the numerically calculated focus measure extremum to the actual distance from focus. This feature allows focusing of the sample in a quick and automatic fashion, to the desired imaging plane, obviating the need for additional auto-focusing apparatuses, and time consuming mechanical focus scans.

A method of performing holographic optical focusing on a plurality of particles in a sample chamber, includes: illuminating a sample of particles in a transparent sample chamber using a coherent light source of an imaging apparatus; acquiring images of said particles using a focusing camera; displaying images of an out-of-focus diffraction pattern of said particles on a display; performing numerical focusing of an imaged hologram of one of said images using a processor of a computer system, to determine a focal plane of said particles; wherein said numerical focusing includes a propagation of said out-of-focus image to different distances which allows a focus measure to be determined numerically by said processor; associating said focus measure with each numerically propagated image, using said processor, such that an extremum in said focus measure with each numerically propagated image can be found; and allowing said computer system to perform a single stage movement of said sample chamber to position said sample in a required focal position.

In one embodiment consistent with the present invention, a method of measuring a particle's affinity to the surface through a measurement of its mobility includes applying a physical force to a plurality of particles disposed on a surface using a physical force application means; and measuring a response of the plurality of particles to the physical force; wherein the measuring step includes one of impulse response measurement or frequency response measurement.

In one embodiment consistent with the present invention, a method of measuring the mobility of particles includes the application of a physical force to a plurality of particles using a physical force application means; measuring a response of the particles to the physical force; wherein the measuring step includes acquiring a sequence of holographic images of the particles in a field of view; and performing a statistical analysis of the sequence of the holographic images synchronized with the application of the physical force to the plurality of particles, to yield a measurement of the mobility of the particles.

In one embodiment, the particles are disposed on a surface, and the physical force application means includes a translation stage that may move the sample, wherein a translation stage movement of sufficient acceleration causes additional particle motion with respect to the surface, and wherein the movement is one of abrupt (e.g. step-like, impulse response measurement) or continuous motion (e.g. frequency response measurement).

In one embodiment, the physical force application means includes optically generated forces using optical forcing means. Alternatively, the physical force application means includes external means, the external means including one of an ultrasonic means, acoustic means, physical probe contact with beads/cells, physical motion of the stage, or an automated fluidic flow device which provides a fluidic flow.

In one embodiment, a method of determining interactions between a plurality of particles and a surface of a sample holder, includes: applying a physical force to at least one of a sample of particles on the sample holder, or to the surface of the sample holder, using a physical force application means; illuminating said particles using an illumination source of an imaging apparatus having a microscope with a field-of-view; measuring a response of said particles to said physical force application means by acquiring a sequence of holographic images of said particles in said field-of-view using said imaging apparatus, said acquisition of said images being synchronized with said physical force application means; statistically analyzing, using a processor of a computer system, said holographic images of said particles captured by said imaging apparatus; wherein said images include a first component that is diffracted by said particles, and a second component that is undiffracted by said particles, and said two components interfere in an imaging plane, yielding an interference pattern produced by said processor, that represents particle by particle intensity fluctuation values; and wherein said particles that are able to move through one of said physical force application means or diffusion, display high intensity fluctuations, and those that are bound to the surface of the sample holder, display low intensity fluctuations, yielding a nature of the interactions on the surface of the sample holder.

In one embodiment, the sequence of images of the sample of particles are captured using a holographic microscope apparatus; wherein each of the images exhibits an interference pattern representing the sample of particles, the interference pattern including a diffracted component and an undiffracted component; and adjusting an amount of defocus of the images to improve a signal-to-noise ratio of the interference pattern. The signal-to-noise ratio is improved when the interference pattern has a higher contrast. The sequence of images is then processed, yielding a statistical image(s) yielding the particle dynamic information. The statistical image(s), which reflect intensity fluctuations of the pixels in the sequence of images of the particles that respond to the physical force, may then be masked with a masking image comprised of particle neighborhoods. The particle neighborhood mask image may be generated by applying standard image processing techniques (e.g. background correction, edge detection, image filters) to the image(s). One frame in the sequence of acquired images is sufficient to determine the neighborhoods of the particles, since the change in the particle positions over the time duration of the image sequence is significantly less than the average interparticle distance. This means that particles may be unambiguously identified with the particle neighborhoods generated using only one image from the sequence. Upon multiplying the statistical image(s) with the particle neighborhood mask, a list of statistical quantities associated with each particle may be generated. These quantities may be averaged over each particle to generate a statistical measurement of particle mobility or particle movement that in aggregate, yield a histogram distribution of such quantities for the field of view. Such distributions may be generated with particles from partial fields of view as well as multiple fields of view as well.

In one embodiment, the statistical image is generated by calculating the pixel-wise standard deviation of the pixel intensities, divided by the pixel-wise average value of the pixel intensities, yielding a statistical image of the normalized standard deviation of pixel intensities for the sequence of images. The statistical image may be masked with an image of particle neighborhoods in order to generate normalized standard deviation values averaged over each particle. The average normalized standard deviation values so generated for each particle may be plotted in a histogram to generate a distribution of values. Particles with high normalized standard deviation are particles that demonstrate high mobility or large movement relative to particles with low normalized standard deviation values. A threshold value of normalized standard deviation may be selected to distinguish the minimally moving fraction of particles (i.e., low normalized standard deviation values) from the particles that demonstrate more freedom of movement (i.e., higher normalized standard deviation). Other threshold values of normalized standard deviation may be applied to select other fractions with intermediate normalized standard deviation values.

In one embodiment, assays involving the measurement of the fraction of particles bound may be performed by measuring the fraction of particles with a value of the normalized standard deviation that is less than the threshold value for bound particles. For a statistically significant fraction of particles to be bound, the background signal level must be characterized. The background signal may be characterized by measuring a similar dispersion of particles that nominally should not be bound under the experimental conditions. The fraction of such unbound particles that are found to be bound (i.e., the fraction of such particles that have a normalized standard deviation value less than the threshold value) under such conditions is the background signal. Knowing this fraction, as well as the number of particles tested in an experiment, allows the application of the binomial probability distribution to determine the statistical probability of generating a given experimental bound fraction measurement from a background source. A minimum threshold of fraction of particles bound may be determined, based on the desired statistical significance of the results, in order to decide whether a positive reaction between the particle dispersion and the treated surface has occurred.

In one embodiment, a calibration curve may be generated using either controlled particle movement calibration experiments, or simulations of controlled particle movements that generate a relationship between the particles' physical motion (e.g., root mean squared distance travelled) and its normalized standard deviation values.

In one embodiment, a statistical image(s) is generated by some combination of pixel statistical measures including average pixel value, pixel standard deviation, pixel variance, higher order pixel fluctuations, pixel temporal correlation functions, pixel spatial correlation functions, pixel spatio-temporal correlation functions, background pixel value, background pixel standard deviation, background pixel variance, higher order background pixel fluctuations, background pixel temporal correlation functions, background pixel spatial correlation functions, background pixel spatio-temporal correlation functions. The pixel statistical measures may be generated pixel-wise over the image sequence and then averaged over the particle neighborhood. The pixel statistical measures may be calculated over the particle neighborhood, and then calculated over the corresponding neighborhoods in the other images in the sequence. The particle neighborhood mask may be generated using one frame in the sequence. Multiple frames from the sequence may be used to generate particle neighborhood masks. Pixel statistical measures may be calculated over subsets of the image sequence. Pixel statistical measures may be calculated over successive subsets of the image sequence generating time varying statistical measurements per particle. Time varying statistical measurements per particle may be associated with time varying experimental conditions (e.g., physical movement, vibration, solution conditions, flow conditions, other environmental effects). Spatial and temporal correlations of particle-based statistical measurements may be performed. Time varying spatial and temporal correlations of particle-based statistical measurements may be calculated (as opposed to pixel-based statistical measurements). Thresholds in the statistical particle measurements may be chosen to select fractions with desired surface affinity, interaction characteristics that have commercial, diagnostic relevance. Thresholds in selected fractions may be chosen to indicate the minimum fractional level necessary to be measured before a positive result is indicated, based on the desired statistical significance.

In one embodiment, particle positions are tracked over time, and statistical measures of particle positions may be generated (e.g., mean squared displacement, net displacement etc.) and distributions of these quantities plotted for multiple particles. Thresholds in particle position measures and particle statistical quantities based on particle positions may be applied to determine fractions with target particle-surface affinities. Statistical measures of particle movement may be based on mean squared particle displacement, mean particle displacement, net particle displacement, higher-order particle position statistics, or a combination of any or all of these quantities. Similar measures of particle movement may be measured for control purposes (e.g., background correction).

In one embodiment, a method of determining interactions between a plurality of particles and a surface of a sample holder, includes: illuminating a sample of particles disposed on a transparent bottom surface of a fluidic flow device, using an illuminating source of an imaging apparatus having a microscope with a field-of-view; measuring a movement of said particles at thermal equilibrium by acquiring a stack of holographic images of said particles in said field-of-view using said imaging apparatus; statistically analyzing, using a processor of a computer system, said holographic images of said particles captured by said imaging apparatus; wherein said statistical analysis includes determining each pixel position through said stack of holographic images, to determine each pixel's standard deviation and its average pixel value; generating a holographic fluctuation image of each said pixel, using said processor; wherein said holographic fluctuation image is a representation of a normalized standard deviation distribution for each said pixel, for said stack of images; processing said holographic fluctuation image, using said processor, to generate a distribution of average normalized standard deviations over each of said particles; wherein relatively larger fluctuations in signal intensity indicate said particles are moving, and relatively smaller fluctuations or a lack of fluctuations in signal intensity indicate said particles are not present; and thereby yielding information about a mobility of said sample and the interaction of said particles on the surface of the sample holder.

In one embodiment, fluctuations of motion of the particles are analyzed at thermal equilibrium, and statistical analysis includes capturing the sequence of images of said sample of particles using a holographic microscope apparatus; analyzing the sequence of images to determine a standard deviation and an average value of each pixel; generating a holographic fluctuation image whereby each pixel value is equal to a value of the standard deviation of the pixel over time, divided by the average value of the pixel; and processing the holographic fluctuation image to generate a distribution of average normalized standard deviations (NSD) over each of the particles.

In one embodiment, the motion of unbound or partially bound particles is more confined at a lower temperature than at a higher temperature. In one example, plots of normalized standard deviation distributions of 4.8 μm silica beads diffusing on a plane glass coverslip taken at three different temperatures, shows that the average value of the distribution increases with temperature, reflecting the increased mean squared displacement of the beads as a function of temperature.

For free diffusion, the mean squared displacement is linearly proportional to the temperature:

Δx²=4Dt, D=k_BT/(6πηr),

where D is the diffusion coefficient (or effective diffusion coefficient), t is the time interval between particle position measurements, k_B is Boltzmann's constant, T is the temperature in Kelvin, η is the viscosity (or effective viscosity) and r is the radius of the spherical diffuser. A comparison between the measured and simulated bead motion values confirms the choice of the normalized standard deviation as an excellent measure of bead mobility. The formula for free diffusion noted above, may also be applied to beads/cells that demonstrate hindered diffusion, generating effective diffusion coefficient estimates.

In one embodiment, the statistical analysis includes capturing the sequence of images of the particles (i.e., red blood cells) using a holographic microscope apparatus; analyzing the sequence of images to determine a standard deviation and an average value of each pixel; generating a holographic fluctuation image whereby each pixel value is equal to a value of the standard deviation of the pixel over time, divided by the average value of the pixel; and processing the holographic fluctuation image of normalized standard deviation values to generate a distribution of average normalized standard deviations (NSD) over each of the particles.

In one embodiment, a method of determining interactions between a plurality of particles and a surface of a sample holder, includes: illuminating a sample of particles disposed on a transparent bottom surface of a fluidic flow device, using an illuminating source of an imaging apparatus having a microscope with a field-of-view; measuring a movement of said particles at thermal equilibrium by acquiring a stack of holographic images of said particles in said field-of-view using said imaging apparatus; statistically analyzing, using a processor of a computer system, said holographic images of said particles captured by said imaging apparatus; wherein said statistical analysis includes determining each pixel position through said stack of holographic images, to determine each pixel's standard deviation and its average pixel value; generating a holographic fluctuation image of each said pixel, using said processor; wherein said holographic fluctuation image is a representation of a normalized standard deviation distribution for each said pixel, for said stack of images; processing said holographic fluctuation image, using said processor, to generate a distribution of average normalized standard deviations over each of said particles; wherein relatively larger fluctuations in signal intensity indicate said particles are moving, and relatively smaller fluctuations or a lack of fluctuations in signal intensity indicate said particles are not present; and thereby yielding information about a mobility of said sample and the interaction of said particles on the surface of the sample holder.

In one embodiment a method of determining an interaction between particles in a sample holder, includes: (1) illuminating a sample of particles disposed on a sample holder, using an illuminating source of an imaging apparatus having a microscope with a field-of-view; (2) measuring a shape of said particles at thermal equilibrium, using a processor of a computer system, by acquiring a sequence of holographic images of said particles in said field-of-view using said imaging apparatus; (3) detecting each of said particles in each of said holographic images using said processor; (4) establishing that a pair of said particles that are adjacent to one another in each of said holographic images, may be bound together and may demonstrate correlated motion, using said processor; (5) extracting an image of each of said adjacent pair of particles, using said processor, creating two sub-images; (6) multiplying said two sub-images together, using said processor, to yield a product sub-image; (7) repeating steps (3)-(6) for all pairs of adjacent particles in all of said holographic images, resulting in a sequence of product sub-images corresponding to each of said pair of adjacent particles; (8) calculating, using said processor, a pixel-wise standard deviation of said sequence of product sub-images; (9) generating an average standard-deviation for one of said pair of adjacent particles by calculating, using said processor, an average value of said pixel-wise standard-deviation divided by an average value for each pixel over a whole product sub-image; wherein adjacent pairs of particles that are bound together have correlated motions which increase a corresponding normalized standard deviation of product sub-image values, than those adjacent pairs of particles that are not bound together, to yield said normalized standard deviation of product sub-image values that are relatively lower with relatively narrower distribution, than unbound particles which exhibit uncorrelated motion, wherein said normalized standard deviation of product sub-image values are relatively higher with relatively broader distribution, such that bound pairs of particles are distinguished from unbound pairs of particles.

In one embodiment consistent with the present invention, the statistical analysis includes: capturing the sequence of images of the sample of particles as holographic images, using a holographic microscope apparatus; detecting each of the particles in each of the holographic images; determining pairs of adjacent particles in each of the holographic images; analyzing each of the pairs of particles by extracting an image of each particle in each of the pairs, creating two sub-images; multiplying (or adding) each of the two sub-images together to yield a product (summed) sub-image; repeating the analysis step for all of the pairs of the adjacent particles; repeating the analysis step for all of the holographic images in the sequence, resulting in a sequence of product (summed) sub-images corresponding to each unique pair of adjacent particles; computing a pixel-wise standard deviation and pixel-wise average arrays of the sequence of the product (summed) sub-images for each of the pairs; calculating an average value of the pixel-wise standard deviation divided by an average value for each pixel over the product (summed) sub-images, to yield an average normalized standard deviation over each of the particles.

In one embodiment consistent with the present invention, the particles/cells are disposed on a surface treated with an antibody, such that the antibody will selectively bind the particles/cells coated with a predetermined antigen. Further, motion of the particles/cells is restricted based on binding of the particles/cells with the antibody on the surface. The particles/cells which do not react with the antibody are unbound and diffuse freely (relatively freely) on the surface.

In one embodiment consistent with the present invention, the particles are red blood cells, and the red blood cells are bound to antibodies on the surface that are specific to antigens on the red blood cells' surface.

In one embodiment consistent with the present invention, the particles are cells which may be expressing surface antigens (e.g., surface receptors), and the cells that are bound to specific antibodies and/or receptor ligands affixed to the surface, have their mobility reduced. Thus by measuring the mobility (e.g., normalized standard deviation), the presence, absence and/or degree of antigen coverage on a cell surface may be determined.

In one embodiment consistent with the present invention, the particles are cells/beads with surface antigens that may be in competition with freely diffusing antigens for surface binding sites. The presence of such diffusing species may affect the mobility of the bound cells/beads in a concentration-dependent way. This type of measurement may be used to determine the presence, absence and/or concentration of freely diffusing species that are in competition with the bead/cell bound species for surface binding sites.

In one embodiment, a method of selective detection of different types of particles on a surface of a sample holder, includes: introducing a sample of particles in a solution onto an antibody coated surface of the sample holder, said particles being coated with either one or another type of antigen; wherein particles coated with one type of antigen are specifically bound to immobilized specific antibodies coated on the sample holder, thereby restricting a motion of said particles; wherein particles coated with another type of antigen are not specifically bound to the sample holder, and said particles freely diffuse in said solution on the surface of the sample holder; illuminating said sample of particles disposed on the sample holder, using an illuminating source of an imaging apparatus having a microscope with a field-of-view; measuring a movement of said particles at thermal equilibrium by acquiring a stack of holographic images of said particles in said field-of-view using said imaging apparatus; statistically analyzing, using a processor of a computer system, said holographic images of said particles captured by said imaging apparatus; wherein said statistical analysis includes determining each pixel position through said stack of holographic images, to determine each pixel's standard deviation and its average pixel value; generating a holographic fluctuation image of each said pixel, using said processor; wherein said holographic fluctuation image is a representation of a normalized standard deviation distribution for each said pixel, for said stack of images; processing said holographic fluctuation image, using said processor, to generate a distribution of average normalized standard deviations over each of said particles; and wherein said specifically bound particles in said field-of-view exhibit relatively lower magnitude and relatively narrower distributions in a normalized standard deviation measure with relatively lower average value, and freely diffusing particles exhibit a relatively higher magnitude and relatively wider distribution with a substantially relatively higher average value of said normalized standard deviation distribution; thereby determining said one or another type of antigens on said particles or said specific antibodies on the sample holder.

In one embodiment consistent with the present invention, the particles are cells/beads with surface antigens which may be bound to a diffusing moiety present in the solution. Furthermore this moiety may simultaneously be able to bind to an appropriately treated surface (solid-phase). In this manner, the presence or quantity of the moiety in solution may be measured by measuring the mobility of appropriately coated particles on appropriately treated surfaces (solid-phase). The presence of such diffusing species may affect the mobility of the bound cells/beads in a concentration-dependent way. This type of measurement may be used to determine the presence, absence and/or concentration of the freely diffusing target moiety which acts as a capture agent for the particles.

In one embodiment, the blood cells not bound to antibodies on the surface demonstrate a broad distribution of the normalized standard distributions, with a high average value of the normalized standard deviation, and blood cells which are bound to antibodies on the surface demonstrate narrow distribution and low average value of the normalized standard distributions.

In one embodiment, intermediate levels of binding are detectable, as indicated by an increased fraction of blood cells with intermediate normalized standard deviations.

In one embodiment, particles with heterogeneous binding properties, detectable by measurement of heterogeneous effective diffusional properties are detectable by an analysis of the width and shape of the distribution, and fitting experimentally measured distributions to heterogeneous diffusion models, yielding the estimates of the population's distribution of diffusional properties which may reflect the distribution of particle-surface interactions and affinities.

In one embodiment, correlated fluctuations occur when there are in-phase motions of the pairs of the particles that are bound together. Further, adjacent pairs of the particles that are bound together will have correlated fluctuations which will increase a value of a corresponding normalized standard deviation of the product (summed) sub-image, in comparison to those that are not bound together and which have relatively lower normalized standard deviations due to uncorrelated fluctuations.

In one embodiment, a method of detection of particle shape as a diagnostic procedure, includes: illuminating a sample of particles disposed on a sample holder, using an illuminating source of an imaging apparatus having a microscope with a field-of-view; measuring a shape of said particles at thermal equilibrium by acquiring a sequence of holographic images of said particles in said field-of-view using said imaging apparatus; statistically analyzing, using a processor of a computer system, said holographic images of said particles captured by said imaging apparatus; wherein said statistical analysis includes determining a spatial intensity of said particles over said sequence of holographic images; generating a holographic fluctuation image of each said pixel, using said processor; wherein said holographic fluctuation image is a representation of a normalized standard deviation distribution for each said pixel, for said sequence of images; processing said holographic fluctuation image, using said processor, to generate a distribution of average normalized standard deviations over each of said particles; wherein said particles that change their shape more readily due to one of thermal fluctuations or externally applied forces demonstrate relatively higher spatial intensity fluctuations compared to said particles that are relatively more rigid; and wherein such normalized standard deviation distributions of said statistical analysis of said spatial intensity of said particles over said sequence of holographic images may be used as a diagnostic of at least one of particle flexibility, elasticity/visco-elasticity, health or disease, age, solution conditions, or binding state.

Thus has been outlined, some features consistent with the present invention in order that the detailed description thereof that follows may be better understood, and in order that the present contribution to the art may be better appreciated. There are, of course, additional features consistent with the present invention that will be described below and which will follow the subject matter of the features appended hereto.

In this respect, before explaining at least one embodiment consistent with the present invention in detail, it is to be understood that the invention is not limited in its application to the details of construction and to the arrangements of the components set forth in the following description or illustrated in the drawings. Methods and apparatuses consistent with the present invention are capable of other embodiments and of being practiced and carried out in various ways. Also, it is to be understood that the phraseology and terminology employed herein, as well as the abstract included below, are for the purpose of description and should not be regarded as limiting.

As such, those skilled in the art will appreciate that the conception upon which this disclosure is based may readily be utilized as a basis for the designing of other structures, methods and systems for carrying out the several purposes of the present invention. It is important, therefore, that the features be regarded as including such equivalent constructions insofar as they do not depart from the spirit and scope of the methods and apparatuses consistent with the present invention.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1A is a schematic diagram, showing an in-line holographic microscopy apparatus designed to measure the mobility of cells/particles in order to infer surface interactions and/or collective diffusion/visco-elastic properties of the cell/particle dispersion.

FIG. 1B is a cross-sectional view of a sample holder having a transparent surface thereon, on which is provided at least one cell/particle.

FIG. 1C is cross-sectional view of a transparent sample holder having a glass slide, spacer and treated coverslip, forming a sample chamber containing a dispersion of particles. The particles may be imaged using the illumination source, microscope objective, and imaging optics and acquisition hardware of FIG. 1A.

FIG. 1D is a cross-sectional view of a sample holder having a reflective/partially reflective surface on lowermost surface (1) or on uppermost surface (2) of a sample holder, on which is provided at least one cell/particle, allowing measurements to be performed in a reflection mode as well. The sample chamber includes a dispersion of particles which settles to a treated coverslip surface.

FIG. 2 illustrates how the hologram of FIG. 1B shifts due to diffusion and may cause large fluctuations in pixel intensity.

FIG. 3A shows a cross-section of a sample chamber with a sample fluid moving therethrough.

FIG. 3B shows a top-down view of differentially treated surface areas (A, B, C and D) of a coverslip which each have a heterogeneous mixture of particles, where each type of particle is distinguishable (by fluorescent label or holographic image characteristics for example).

FIG. 4, screenshot A shows a holographic image of 2 μm silica beads, and screenshot B shows a fluctuation image generated from a stack of holographic images (60 images).

FIG. 5A shows an analysis of a stack of holographic images, yielding the holographic fluctuation image S(i, j) which is constructed by dividing the pixel-wise standard deviation by the pixel-wise average, for each pixel, over the stack of images.

FIG. 5B shows how per-particle fluctuation measurements are calculated using the fluctuation image (i.e., normalized standard deviation (NSD) image) that is masked with a particle neighborhood masking image generated from one image in the sequence of holograms. The normalized standard deviation values over each particle neighborhood are averaged, and then plotted as a histogram.

FIG. 6, screenshot A, shows a holographic fluctuation image of 2 μm silica beads in deionized water at 5° C. (constructed from a 60 frame stack), and screenshot B shows a holographic fluctuation image of 2 μm silica beads in deionized water at 47° C. (constructed from a 60 frame stack). The light areas, which represent the high intensity fluctuation areas due to particle movement, are more confined at the lower temperature.

FIG. 7 shows three histograms of normalized standard deviations of 4.8 μm diameter silica beads in water measured at 6.8° C. (lower histogram), 28° C. (middle histogram), and 46.3° C. (top histogram). The main peaks in the histograms correspond to the freely diffusing beads, while the minor peak at lower values of NSD (5-10%) represent the much smaller fraction of beads that are stuck or partially stuck to the surface.

FIG. 8 is a plot of the mean and standard deviation of NSD values for freely diffusing 4.8 μm beads measured at various temperatures (open squares).

FIG. 9, slide (1), shows each type of particle is placed on an antibody coated surface (i.e., direct immunoassay). The particle coated with A antigen is specifically bound to the immobilized specific antibodies (A antibodies) coated on the coverslip surface, thereby restricting the motion of the particle.

FIG. 9, slide (2), shows the differing behavior of each type of particle, where the B-antigen coated particle which is not specifically bound, is allowed to freely diffuse in the solution, on the surface, while the A type particle motion is severely restricted due to specific binding with the surface antibodies.

FIG. 10A shows plots of normalized standard deviation histograms of two different samples or RBCs (red blood cells) measured on an anti-A coated surface. Red blood cells with A antigen on the surface (i.e., A type) specifically bind the anti-A coated surface, while blood cells without the A antigen (i.e., A negative) do not bind, and are thus, free to diffuse.

FIG. 10B shows plots where the specifically bound cells that are positive for antigen A (type A, black histogram) have a similarly narrow distribution with similarly low average value, compared to the A positive histogram of FIG. 10A.

FIG. 11, slide (1), shows each type of particle is placed on an antigen coated surface that is in the presence of a complementary antibody in solution (i.e., indirect immunoassay). The A antibody in solution is able to bind to the A antigen coated surface, as well as simultaneously bind to a particle coated with A antigen.

FIG. 11, slide (2), shows the differing behavior of each type of particle, where the B-antigen coated particle which is not specifically bound, is allowed to freely diffuse in the solution, on the surface, while the A type particle motion is severely restricted due to specific binding with the antibody that is itself specifically bound to the surface.

FIGS. 12A and 12B show a series of NSD histograms for two samples of red blood cells taken at different times. FIG. 12A includes type A red blood cells in the presence of a high concentration of anti-A antibody (100 nM) dispersed in synthetic plasma onto a surface with type B antigens on a coverslip surface. The cells are able to diffuse, as evidenced by the high average values of the NSD observable. FIG. 12B shows a time series of histograms under similar conditions to those in FIG. 12A, except that the surface had type A antigens on it (unlike the type B antigen surface measurements of FIG. 12A). Comparison of the histograms of FIGS. 12A and 12B allows us to choose a threshold value for NSD to determine whether a cell is bound or not.

FIGS. 13A and 13B are similar to the conditions measured in FIGS. 12A and 12B except that a much lower concentration of antibody, 1 nM anti-A, was used in the synthetic plasma. In FIG. 13A, the A type cells on the B type surface are mostly unbound when first measured at 19 minutes after the sample was introduced onto the surface. In FIG. 13B the A type cells in synthetic plasma containing 1 nM anti-A were measured after being introduced onto a surface with A type antigens.

FIG. 14 shows the cell adhesion kinetics for the same sample studied in FIGS. 13A and 13B.

FIG. 15, slide A, shows an image of a 4.8 μm diameter silica bead, where a sequence of 40 frames of this diffusing bead (acquired at 5 frames per second) was analyzed using a particle tracking algorithm which yielded a sequence of shifts in the measured centroid position, Δx and Δy, as plotted in FIG. 15, slide B.

FIG. 16, slide A, shows the image of the NSD calculated from the same measured sequence of 40 frames that was used in the particle tracking analysis shown in FIG. 15, slide B, which yielded a NSD value, averaged over the bead area, of 23.6% NSD. The averaged NSD value over the bead area in each member of the ensemble is plotted in FIG. 16, slide B, showing excellent agreement with the experimentally measured NSD.

FIG. 17 shows a calibration curve relating the mean of the measured bead NSD distribution to root mean squared displacement.

FIG. 18 shows the calibration curve plotted in FIG. 17, as well as two other calibration curves generated from two other bead images (4.8 μm diameter silica).

FIG. 19 shows the beads from FIG. 18, along with a quantity termed the contrast correction factor. This factor is determined by calculating the standard deviation of the pixel intensities in the neighborhood of each bead (i.e., within rectangle surrounding each bead), divided by the mean value of the pixel intensities in the same region (i.e., normalized spatial standard deviation).

FIG. 20, plot A, is an NSD histogram of B type red blood cells in synthetic plasma which includes anti-B, dispersed on a surface with A antigens. FIG. 20, plot B, is a scatter plot that relates the NSD value of each red blood cell measured in FIG. 20 plot A, to its normalized spatial standard deviation.

FIG. 21, plot A, shows an NSD histogram of B-type red blood cells in synthetic plasma which includes anti-B, dispersed on a surface with B antigens. The anti-B antibodies attach to the B antigens on the red blood cells, as well as the B antigens on the surface, thereby immobilizing the cells. As a consequence the intensity fluctuations generated are significantly reduced compared to the cells on the A antigen patch (FIG. 20, plot A), as is shown by the lower NSD measurements in FIG. 21, plot A. FIG. 21, plot B, is a scatter plot that relates the NSD value of each the red blood measured in FIG. 21, plot A, to its normalized spatial standard deviation.

FIG. 22, plots A and B, show corrected and uncorrected histograms of unbound (FIG. 20, plot A) and bound (FIG. 21, plot A) red blood cell NSD histograms, where corrections were applied by using the linear fits (FIG. 20, plot B and FIG. 21, plot B) to remove the NSDs dependence on the normalized spatial standard deviation.

FIG. 23 is a plot of the cumulative probability distribution of the histograms shown in FIG. 22, plot A, and FIG. 22, plot B (i.e., a plot of the fraction of cells that have an NSD less than or equal to a particular NSD value).

FIG. 24 shows an NSD histogram of 4.8 μm diameter beads in water (top histogram) as well as a simulated NSD histogram using a Gaussian distribution that yields a square-root mean squared displacement of 0.131 μm (bottom histogram).

FIG. 25 shows an NSD histogram of 4.8 μm diameter beads in water with 0.4% saline (top histogram) as well as a simulated NSD histogram using a Gaussian distribution that yields a square-root mean squared displacement of 0.0064 μm (bottom histogram).

FIG. 26, plots A, B, and C, show NSD histograms of type A red blood cells in synthetic plasma containing 5 nM anti-A dispersed on a surface with B antigen.

FIG. 27, plots A and B, are plots of the normalized spatial standard deviation values for unbound and bound red bloods respectively measured from one image.

FIG. 28, plots A and B, plot the average spatial normalized standard deviation for 100 cells in the unbound population of FIG. 20, plot A, and FIG. 24, plot A (FIG. 28, plot A), as well as that for 100 cells in the bound population of FIG. 20, plot B, and FIG. 24, plot B (FIG. 28, plot B) over all 40 frames in the sequence. In both of the plots the error bars plot the standard deviation of each cell's spatial normalized deviation over the 40 frame sequence.

FIG. 29 plots the standard deviation of the spatial normalized standard deviation measured for each cell, over the 40 frames in the sequence for both the bound cells as well as the unbound cells.

FIG. 30 shows a calibration curve for using holographic focusing. The calibration sample containing particles is displaced known distances from the focal position, and an image is acquired at each distance. Each image is then numerically propagated and the position of its focal measure extremum is plotted as a function of sample distance from focus.

FIG. 31 shows the final position of the sample that initially started in a range of initial positions between +1 mm and −1 mm away from focus. The in-focus position was determined visually and has an error estimated as ±1 μm. The accuracy of the holographic focusing is comparable to the error in the focal position estimation.

FIGS. 32A and 32B show a correlation figure for pairs of particles/cell where sub-images are used to distinguish bound particle/cell pairs from unbound particle/cell pairs.

FIG. 33A shows bead intensity fluctuations (NSD, normalized standard deviation values) of two different samples of bead dispersions (4.8 μm silica) measured without the application of any external forces (i.e., a thermal equilibrium measurement), where one sample was bound to the surface (dense cross-hatched histogram) and the other was unbound to the surface (histogram with sparse diagonal line pattern). The beads were made to bind to the surface of a glass coverslip by diluting them in a 0.9% saline solution, while the beads were made to freely to diffuse on the coverslip by dilution in deionized water. The unbound beads are distinct from the bound ones under these experimental conditions (i.e., minimal overlap of distributions). FIG. 33B shows the same samples measured after the application of stage movements. Each image in the stack of analyzed histograms was taken after a piezo-electric stage moved the sample 25 μm back and forth along one axis. This physical perturbation on the system did not affect the mobility of the bound population of beads (dense cross-hatched histogram) in comparison to the measurements taken without force application, as measured by the NSD measurements. However the unbound beads displayed higher NSD values than their non-forced counterparts in FIG. 33A as a result of the increased bead mobility generated by the stage motions. Physical force application can thus better resolve bound and unbound populations as evidenced by the greater separation between the bound and the unbound bead histograms in FIG. 33B.

DESCRIPTION OF THE INVENTION

The present invention relates to an instrument and measurement apparatus and methodology used for measurements and testing, that characterize a population of cells/particles or which can detect a sub-population of cells/particles, based on their detected mobility, in a quick and efficient manner.

Apparatus

In one exemplary embodiment, FIG. 1A shows a schematic diagram of an in-line holographic microscope apparatus 10 designed as the means to measure the mobility of particles (i.e., cells) in order to infer surface interactions (i.e., presence or absence of specific surface—particle interactions), and/or collective diffusion/visco-elastic properties of the particle dispersion (e.g., effective viscosity).

In the exemplary embodiment of FIGS. 1A and 1C, the apparatus 10 used for carrying out the present invention includes a coherent light source (e.g., laser, superluminescent diode) 100 that is collimated by a collimator 101, which may be coupled to an optical-fiber 103, and whose laser beam 104 illuminates a transparent sample holder 105 (i.e., microscope slide) of a microscope 106.

In one embodiment, the coherent light source 100 is a laser which has a short coherence length (<400 μm), and operates at 660 nm. Other wavelengths and types of illumination sources, including non-laser sources (e.g., superluminescent diodes) and conventional light sources (e.g. LED, incandescent, arc lamp) may be used as well. Holographic optical trapping apparatuses are well known in the art, as disclosed in U.S. Pat. No. 7,161,140, to Grier et al., for example, which is herein incorporated by reference in its entirety.

In one embodiment, the sample 107 disposed on the sample holder 105, includes a dispersion of particles 108 (i.e., cells) disposed on a treated or untreated transparent surface 109 (i.e., coverslip 109A). The sample 107 of cells/particles 108 may be introduced into the sample holder 105 manually or through an automated fluidic device 116 (discussed later), the structure and operation of which is well known in the art.

The particles 108 may settle on the surface 109 to test particle-surface interaction. In one embodiment, the particles 108 may settle on the surface 109 due to gravitational forces. In another embodiment, the particles 108 may settle on the surface 109 due to centrifugal forces applied to a sample chamber 118 (see FIG. 3A) using a centrifuge (117, for example). Further, in another embodiment, the particles 108 may settle on the surface 109 due to other forces applied to the particles 108 (discussed further herein).

The particles 108 used may have a variety of physical and chemical attributes, and may be of different types, based on size, shape, and materials, yielding distinguishable images on the image formation apparatus. For example, a particle 108 may be a regularly shaped bead with some symmetry (e.g., spherical, prolate spheroid, oblate spheroid), or an irregularly shaped bead. A particle 108 may be made of one type of material or of multiple types of materials. A particle 108 may be solid, porous, or have a hollow core. A particle 108 may be fully or partially coated with other material(s). A particle 108 may be metallic or partly metallic. A particle 108 may be non-metallic or partly non-metallic. A particle's 108 surface may be treated to apply a texture. A particle 108 may be a silica bead with linker molecules on the surface, or a silica bead with biomolecules or synthetic molecules attached to the surface. A particle 108 may also be a silica bead with biomolecules or synthetic molecules attached to the linker molecules on the surface. A particle 108 may be a bead coated with or otherwise embedded with a fluorescent or luminescent label molecule(s) covalently or non-covalently attached to it or integrated with it, which may also distinguish particle type based on fluorescent or luminescent emission spectrum. A particle 108 may be a bead coated or embedded with a combination of different fluorescent or luminescent label molecule(s) covalently or non-covalently attached to it or integrated with it, which may also distinguish particle type based on fluorescent or luminescent emission spectrum. A particle 108 may be a bead with a nanoparticle(s), or magnetic nanoparticle(s), or fluorescent nanoparticle(s), covalently or non-covalently attached to it or integrated with it.

In further example, a particle 108 may be a biological cell. A particle 108 may be a genetically engineered biological cell or a descendant of a genetically engineered biological cell. A particle 108 may be a cell that is treated with biomolecules and/or synthetic molecules. A particle 108 may be a cell that is treated with linker molecules. A particle 108 may be such a cell that is treated with biomolecules and/or synthetic molecules that attach to the linker molecules. A particle 108 may be a cell with a fluorescent or luminescent label molecule(s) covalently or non-covalently attached to it or integrated with it. A particle 108 may be a cell with a combination of different fluorescent or luminescent label molecule(s) covalently or non-covalently attached to it or integrated with it. A particle 108 may be a cell with a nanoparticle(s) covalently or non-covalently attached to it or integrated with it. A particle 108 may be a cell with a magnetic nanoparticle(s) covalently or non-covalently attached to it or integrated with it. A particle 108 may be a cell with a fluorescent nanoparticle(s) covalently or non-covalently attached to it or integrated with it. A particle 108 may be a cell that naturally expresses or is genetically altered to express fluorescent protein(s).

In one embodiment, the sample 107 of particles 108 may further be modified by introducing reagents or non-reactive solutions onto the sample holder 105—before, during or after the measurements. For example, as shown in FIG. 1B, for the measurement of the surface interactions of the cells/particles 108, the transparent surface 109 (e.g., coverslip 109A) of the sample holder 105 onto which the cells/particles 108 settle, may be provided with special treatment.

For example, the surface 109 (including coverslip 109A) may be provided in a variety of ways. In one embodiment, the surface 109 may be flat and transparent. Further, a surface 109 may be flat and partially transparent. Still further, a surface 109 may be a flat and fully or partially reflective. A surface 109 may also be a textured flat surface. A surface 109 may be treated with biomolecules or synthetic molecules. A surface 109 may be treated with linker molecules. A surface 109 may be treated with biomolecules or synthetic molecules linked to the linker molecules. A surface 109 may be differentially treated with a variety of molecules. A surface 109 may be differentially treated with a variety of linker molecules. A surface 109 may be treated with a mixture of molecules. A surface 109 may be part of a microfluidic device. A surface 109 may be part of a microtiter plate. A surface 109 may be part of a transparent or partially transparent sample chamber 118 (see FIGS. 1D and 3A). A surface 109 may be part of a reflective or partially reflective sample chamber 118 (i.e., reflective or partially reflective surface 120 of FIG. 1D). A surface 109 may be sufficiently microscopically smooth to allow imaging of particles 108.

In one example, FIG. 3B shows a top-down view of differentially treated surface areas (A, B, C and D) of a coverslip 109A on a sample holder 105 which each have a heterogeneous mixture of particles 108, where each type of particle 108 is distinguishable (by fluorescent label or holographic image characteristics, for example). Note that each surface 109, 109A yields a differing interaction (i.e., different types of particles 108 are able to diffuse in each area) with respect to the mobility measure of the heterogeneous population of particles 108. FIG. 3B demonstrates how many different types of interactions may be probed simultaneously (i.e., multiplexed measurements) by using a distinguishable heterogeneity of particles 108 with or without differentially treated surface 109, 109A areas, in accordance with the present invention.

In one embodiment, a heterogeneous population of particles 108 may be measured simultaneously, with each type of particle testing for different quantities or regimes in similar quantities (e.g., multiplexed measurements) (see FIG. 3B). All or some types of particles 108 may be distinguished based on a label (e.g., fluorescent, nanoparticle), particle image (due to different absorption, scattering, fluorescence, luminescence characteristics, fluorescence or luminescence emission profiles, fluorescent or luminescent decay lifetime), and/or particle position (assuming controlled deposition of particle types). All or some types of particles 108 may be distinguished by a multimodal collection of data of each particle 108. All or some types of particles 108 may be distinguished based on their holograms generated by a coherent light source. All or some types of particles 108 may be distinguished based on their focused and/or defocused holograms generated by a coherent light source 100. All or some types of particles 108 may be distinguished based on their holograms collected at more than one focal plane. A homogenous population of particles 108 may be measured simultaneously.

In one embodiment, the sample 107 includes particles 108 coated with biomolecule A, surface- (i.e., solid-phase) coated with biomolecule B in the presence of solution containing (or not containing) biomolecule A, as well as containing (or not containing) other types of biomolecules (e.g., biomolecule C, D, E, etc.). The sample measurement of this method yields information of biomolecular interactions between the particles 108 and the surface 109 in the presence of the solution for research, industrial and/or clinical purposes by an analysis of particle mobility in response to controlled or thermal forces.

In one embodiment consistent with the present invention, the particles 108 are cells/beads 108 with surface antigens which may be bound to a diffusing moiety present in the solution. Furthermore this moiety may simultaneously be able to bind to an appropriately treated surface (solid-phase). In this manner, the presence or quantity of the moiety in solution may be measured by measuring the mobility of appropriately coated particles 108 on appropriately treated surfaces 109 (solid-phase). The presence of such diffusing species may affect the mobility of the bound cells/beads 108 in a concentration-dependent way. This type of measurement may be used to determine the presence, absence and/or concentration of the freely diffusing target moiety which acts as a capture agent for the particles.

In an exemplary embodiment, if particular antigens are probed on the cell/particle surface 108a (see FIGS. 1B and 1C), an appropriate surface treatment for the transparent surface 109 may include chemically modifying the surface 109 with a suitable linker molecule to be able to attach the appropriate antibody to the surface 109, which allows specific binding of the cell/particle 108 presenting the target antigen with the surface antibody on the transparent surface 109. Cells/particles 108 that do not have the target antigen on their surface 108a do not specifically bind to the surface of the transparent surface 109, in this case. In the case of measuring cell/particle 108 diffusional properties (i.e., effective diffusion coefficient, effective viscosity or visco-elasticity properties), for example (discussed later), an inert surface 109 would be used.

The cells/particles 108 may be imaged by a microscope objective lens 110 (see FIG. 1A) of a microscope apparatus, and a tube lens 111 which allows the magnified pattern of the cell/particles 108 to be imaged on a CCD or CMOS camera 112, which is connected to a computer 113 for image processing etc. Different areas of the sample 107 may be imaged by translating the sample 107 with a translation stage 114 (e.g., motorized microscope translation stage), which is well-known in the art. The sample holder 105 may also allow all or parts of the measured areas to be temperature controlled using a temperature controlling apparatus therein, allowing for example incubations to occur at optimal temperatures for biomolecular interactions to take place. Furthermore, the focal plane that is imaged by the microscope 106 may be adjusted by using the focus control 115 (which may be motor driven as well).

In one embodiment, when the particles 108 are coated or otherwise have embedded fluorescent, or luminescent molecules or nanoparticles, which are distinguished by particle type based on fluorescent or luminescent emission spectrum, the apparatus 10 shall be equipped with a fluorescent, luminescent excitation source 100, appropriate filters (not shown) and dichroic elements (not shown) as well as color detection capabilities (e.g., color camera 113 and/or emission filter selections). The introduction of different types of particles 108 with each type being uniquely coated may allow multiplexing the measurements (i.e., measurement of multiple types of interactions simultaneously).

Unlike in many traditional fluorometric multiplexed measurements, a separate washing step is not needed, since it is not the presence of a given particle 108 that marks a positive binding interaction, it is the mobility measurement of the particle that indicates positive binding.

In another embodiment consistent with the present invention, a transparent, semi-transparent, or partially mirrored sample 107 and sample chamber 118 with reflective coating 120 (see FIG. 1D), may be measured in reflection mode, involving the laser illumination (or alternative illumination source) to illuminate the sample 107 from the collection side, and image formation occurring from the light reflected from the sample 107 that is subsequently imaged onto the computer 113 monitor.

In yet another embodiment consistent with the present invention, the transparent sample holder 105 is an automated fluidic device 116 or microtiter plate device with data acquisition and analysis capability (see FIG. 3A), as described herein. The particles 108 are flowed into the sample chamber 118 of the microfluidic device 116, or introduced into the microtiter plates by a robotic microtiter apparatus which introduces a number of samples 107 (particles 108 and solution) into different wells. Each well may have a unique surface chemistry, indexed by position, allowing multiple tests to be performed, such as independent binding assays, ultimately allowing a single sample 107 (or a plurality of samples 107) to be tested with a plurality of surfaces 109.

The transparent surface 109 on which the particles 108 settle may be made of treated plastic or glass. The microtiter plate may or may not have a customized configuration, including optically transparent caps, sample delivery zones, sample viewing zones etc. The instrumentation outlined herein may be integrated with a robotic microtiter plate handling machine for automated (and parallelizable) fluid delivery from sample containers to each well, sample mixing and incubation capabilities, as well as parallelized microtiter plate measurement capabilities, which may be programmable and automated. Thus, an assay may be designed to perform multiple tests on one or a number of samples 107, in a parallel fashion. In addition, multiple microtiter plates may be measured in a parallel fashion with a suitably parallelized optical train and detection set-up. A robotic apparatus may feed microtiter plates into the detection area for measurement in an automatic fashion, allowing stacks of microtiter plates to be measured without user intervention.

Measurements in a given sample chamber 105, microfluidic sample chamber 105 or microtiter plate well, may be repeated after the addition of solutions, particles or mixtures, or the exchange of solutions, particles or mixtures, and/or incubations at different temperatures.

Titration measurements may be performed in a given sample chamber 105, microfluidic sample chamber(s) 105 or microtiter plate well(s) by introducing additional analytes into the chamber/well solution(s).

Further, kinetic experiments may be performed in a given sample chamber 105, microfluidic sample chamber(s) 105 or microtiter plate well(s) by following the time-course of the particle mobility measurements (e.g. NSD) after the introduction of additional analytes into the chamber/well solution(s) (or not).

The disclosed assay methods may be implemented as a computer program product for use with a computer system. Such implementations may include a series of computer instructions fixed either on a tangible medium, such as a computer readable medium (e.g., a diskette, CD-ROM, ROM, or fixed disk) or transmittable to a computer system, via a modem or other interface device, such as a communications adapter connected to a network over a medium. The medium may be either a tangible medium (e.g., optical or analog communications lines) or a medium implemented with wireless techniques (e.g., microwave, infrared or other transmission techniques). The series of computer instructions embodies all or part of the functionality previously described herein with respect to the system. Those skilled in the art should appreciate that such computer instructions can be written in a number of programming languages for use with many computer architectures or operating systems.

It is expected that such a computer program product may be distributed as a removable medium with accompanying printed or electronic documentation (e.g., shrink wrapped software), preloaded with a computer system (e.g., on system ROM or fixed disk), or distributed from a server or electronic bulletin board over the network (e.g., the Internet or World Wide Web). Of course, some embodiments of the invention may be implemented as a combination of both software (e.g., a computer program product) and hardware. Still other embodiments of the invention are implemented as entirely hardware, or entirely software (e.g., a computer program product).

Binding Techniques

In one embodiment, the plurality of particles 108, such as cells 108 expressing receptors, are disposed on a surface 109 (i.e., solid-phase), treated with an analyte such as a receptor ligand (i.e., solid-phase) such that said ligands will selectively detect some fraction of particles 108 coated with receptors complementary to the ligands.

The particles 108 compete with the target receptor in the sample solution for binding sites on the solid-phase, for the purposes of determining receptor concentration in the sample solution. The solution containing the target receptor may be preincubated on the solid-phase surface 109 before the introduction of the coated particles 108.

In one embodiment, a plurality of particles 108 are cells 108 expressing proteins on their surface, and are disposed on a surface 109 (i.e., solid-phase) which is treated with a moiety such as a protein binding receptor or an antibody such that said surface proteins will selectively be bound by said solid-phase receptors or antibodies. In other words, the particles 108 are coated with complementary antigen, which then competes with antigen in the sample solution for binding sites on protein binding receptor/antibodies immobilized on solid-phase. As noted above, the sample solution with target antigen may be preincubated in a sample chamber 105, microfluidic sample chamber 105 or microtiter well, containing the solid-phase, before introduction of the coated particles 108, which then bind to unoccupied antibody sites on solid-phase, and are distinguishable from unbound particles 108 based on their mobility. With calibration, the method may yield target antigen concentration in the sample solution.

The coverage density of antigens/antibodies/analytes/surface proteins on particles 108 (or expression levels of surface proteins in cells) may be estimated based on the degree of restricted motion observed on a known solid-phase, and with known solution conditions.

Solid-phase consists of immobilized capture antibody specific to a particular epitope(s) of an analyte of interest, and particles 108 are coated with a secondary antibody (i.e., like a “label antibody”) specific to epitopes on different regions of the same analyte such that the solid-phase capture antibody and particle bound “label antibody” may be simultaneously bound to the analyte (i.e., sandwich assay). Particles 108 that bind to the surface 109 due to such an interaction may be used as a measurement of the analyte concentration of the unknown sample 107. In addition, the kinetics of fractional particle binding (i.e., fraction of particles bound) may be used as a method to measure analyte concentration as well as analyte biochemical properties (e.g., rate constants, equilibrium constants etc.). Such a measurement does not require a “label antibody” wash step, since the bound “label antibodies” (i.e., antibody coated particles) may be distinguished from the unbound particles 108 by their diffusive behavior (i.e., bound particles show significantly diminished mobility), or response to physical forces (i.e., bound particles show significantly diminished response to physical force). Unlike in traditional immunometric assays, “label antibodies” do not necessarily indicate a positive binding interaction. Their presence is necessary but not sufficient. The positive binding interaction is finally determined by the particle's mobility measurement.

In one embodiment, the plurality of particles 108 are red blood cells 108, and are bound to the antibody on the surface 109.

In one embodiment, the target antibodies are taken from a blood sample, and testing is done against an array of uniquely treated surfaces to determine an antibody profile. Specifically, the target antibodies are taken from a blood sample for the purposes of detecting viral infection. Proteins that occur on the surface of a given virus may be immobilized on the surface (i.e., solid-phase) thereby being able to capture the specific antibody to that virus. In addition, particles coated with antibodies complementary to another region of the virus antibody are present in the test, such that in the presence of the target virus antibody, immobilization of particles may occur, signaling the presence of the antibody in the blood sample. Such measurements are performed in order to diagnose infection, or quantify target antibody concentration, with suitable controls.

The magnitude of particle binding, and the kinetics of binding strength, as measured by decreased particle mobility with increased particle binding, may be used as a method to measure analyte concentration and biochemical properties (e.g., rate constants, equilibrium constants etc.).

Holographic Focusing

The measurement of multiple fields of view, sample chambers 118 (see FIG. 3A), microfluidic chambers 118 and/or microtiter well plate chambers may require multiple refocusing events. To increase the throughput of such a measurement, focusing time should be kept to a minimum.

Traditional focusing techniques require multiple focus shifts, and a comparison of the images acquired at these different focal positions. By applying a focus measure to each image, a determination may be made as to whether the focus shifts are approaching the true focal position or not. The fact that the sample 107 is physically moved multiple times to determine the focus makes such techniques time consuming.

However, in contrast to traditional techniques, the use of a coherent source 100 to image the particles 108 allows numerical processing of an imaged hologram to determine the focal plane of the particles 108 in a quick fashion. Illuminating the sample 107 of particles 108 in a transparent sample chamber 118, for example, with a coherent source 100, allows imaging of particles' 108 diffraction pattern even when significantly out of focus. A numerical solution to focusing allows quick focusing over a long-range of out-of-focus distances. Numerical focusing involves the propagation of the out-of-focus image to different distances which allows the focus to be determined numerically. By associating a focus measure with each numerically propagated image, an extremum in the focus measure may be found, allowing a single stage movement to position the sample in the required focal position. This is in contrast to the traditional focusing methods which require comparatively slow physical scans of a sample through different focal distances to determine the focal position according to some focus measure.

Thus, the acquired out-of-focus diffraction pattern may be numerically propagated over a range of distances to determine the distance that maximizes the focus measure. Once this distance is numerically determined, a single stage movement may be performed to position the sample 107 to the required focal position.

A dedicated focusing camera 217 (see FIG. 1A) may be used in acquiring images upon which the numerical propagation calculations are performed. In one embodiment, this camera may be placed at an imaging plane different to that of the camera 112 used for mobility measurement by means of a beam splitter 218. Thus, focusing calculations are performed from images collected from this separate focusing camera 217 which is positioned in a plane that is out-of-focus compared to the measurement camera 112.

It is also possible to use a single camera 112 for both focusing as well as mobility measurement. An image for focusing purposes may be collected after controlled defocusing of the sample 107.

FIG. 30 illustrates a calibration curve generated from numerically propagating images form a calibration sample 107 that was displaced from 1 mm above the true focus position to 1 mm below focus. The images were collected on a dedicated focusing camera that was displaced 1 mm away from the focus, in object space. The objective 110 used had a low-magnification and low-numerical aperture objective. Calibration curves may also be generated with different objectives 110 and CCD positions as well. The x-axis of the calibration curve shows the actual distance the calibration sample 107 was moved, with respect to the focal position, and the y-axis value indicates the location of the numerically propagated peak of the focal measure. For samples 107 that are defocused significantly far from the focal position, more than one numerical focus iteration may be required to attain the desired focusing accuracy.

FIG. 31 illustrates the focusing performance using the calibration curve of FIG. 30 when three focusing iterations are performed for a range of initial starting positions ranging from +1 mm to −1 mm. The true focal position was determined visually and has an error of approximately ±1 μm. The values of the final position with respect to the focal-plane are accurate to within the estimated error of the visually determined focal position measurement.

Methods and Apparatuses of Performing the Invention There are a number of methods of performing the present invention with the above-described apparatus—which are described herein below.

Physical Force Application Method

In one embodiment consistent with the present invention, a physical force is applied to the cells/particles 108 and/or surface 108a thereof, using physical force application means 10 (see FIGS. 1A and 1B), and the response of the cell/particle 108 motions can be measured (i.e., impulse response measurement, frequency response measurement) using the computer 113. The physical force application means 10 may include an abrupt or continuously periodic movement of the translation stage 114, or optically generated forces using optical forcing means 100, 101 etc., or by the application of other external means such as an ultrasonic means 117 which uses an ultrasonic emitter 117 which applies ultrasonic waves to the sample 107, acoustic means (i.e. acoustic wave application), or a physical probe contact, or by fluid flow from an automated fluidic flow (microfluidic) device 116, for example, all of which are well-known to one of ordinary skill in the art.

In the physical force application method stated above, the measurement methods of the present invention involve acquiring sequences of holographic images of cells/particles 108 in a field of view (e.g., as viewed through the microscope 106) (see FIGS. 1A, 1B, and 2).

In the case of a measurement done in concert with the application of an external force (i.e., physical force application method), a statistical analysis of a sequence of holographic microscope images synchronized with the physical force application, yields the nature of the interactions on the surface 108a or the diffusional/visco-elastic characteristics of the cell/particle 108 dispersion.

Specifically, in the analysis by the computer 113 of the images of the cell/particles 108 captured by the imaging device 112, the images include a component that is diffracted by the cells/particles 108, as well as an undiffracted component. These two components interfere in the imaging plane, yielding an interference pattern that represents the sample 107. The use of coherent illumination that is spatially coherent allows the sample 107 to be imaged even when it is significantly out of focus. By adjusting the amount of defocus, the interference pattern may be adjusted in order to improve the signal-to-noise ratio of the measurements.

In the exemplary measurements discussed below, the focal distance was adjusted so that the interference pattern had a trough of low intensity in the center (see FIG. 2) with a bright intensity ring surrounding it. Additional concentric dark and bright rings with a lower modulation depth extend away from the center. A high signal-to-noise ratio may be achieved by forming an interference pattern with a low central intensity and a high surrounding ring of intensity.

In this embodiment, beads or cells (i.e. particles) 108 that do not move, mean that the fluctuation of their hologram pixel counts is dominated by the effects from environmental vibrations, photon statistics and detector noise which is a relatively small in magnitude. Particles or cells 108 that do move (e.g., by physical force or diffusion) will have their interference patterns shift as the particles or cells 108 move. When a cell or particle 108 moves in one frame with respect to a previous frame, a pixel with a low count in the central trough in the first frame may be exposed to the adjacent bright ring in the succeeding frame. This large fluctuation of pixel intensity from low count value to high count value, if repeated, will yield an average pixel fluctuation value which is high (see graph in FIG. 2). Normalizing such fluctuations (i.e., dividing the standard deviation of the pixel values by their average values, pixel by pixel), weighs fluctuations in intensity that occur at the center of the particle/cell 108 hologram the highest, since the average values are the lowest in this region.

In summary, this embodiment yields particle by particle intensity fluctuation values, whereby particles 108 that are able to move through physical force application or diffusion, display high intensity fluctuations, and those that are bound to the surface 109 display low intensity fluctuations. This intensity fluctuation measurement is the measurable that is used to determine particle binding, in particle binding based assays.

Unlike in many other types of binding assays, washing of the surface 109 to remove unbound particles 108 is not required since binding is determined based on the mobility measurement, and unbound particles 108 are distinct from bound ones in this regard.

Specifically, FIG. 2 illustrates how the hologram shifts due to diffusion and may cause large fluctuations in pixel intensity. For pixels that are in the central, low intensity portion of the hologram, shifts in the hologram position cause large fluctuations when the surrounding bright fringe of the hologram is detected. Pixel k in the diagram detects the central minima of the hologram when the hologram is in position x₁. A shift of the hologram to position x₂ causes pixel k to see the peak intensity of the surrounding ring. A return of the hologram to the original position x₁ causes pixel k to return to a low count rate again. The overall effect is that the central region of minimal count values picks up large pixel intensity fluctuations by virtue of hologram movements causing the surrounding bright fringe to periodically contribute high pixel intensities.

As stated above, the coherent laser source 100 has a short coherence length (<400 μm), and operates at 660 nm. The short coherence length prevents interference fringes from being formed, which would be due to the coherent superposition of optical reflections with the transmitted beam. As stated above, other wavelengths and types of illumination sources, including non-laser sources (e.g., superluminescent diodes) and conventional light sources (e.g. LED, incandescent, arc lamp). Any source 100 that allows the particles 108 to be imaged with sufficient contrast is usable.

In this embodiment, the CCD 112 is used to capture the magnified images of the sample 107. The exposure time of the CCD 112 (or the pulse duration of a pulsed laser 100, in the case of using a pulsed laser as a strobe source) should be significantly shorter than the diffusional time constant. This ensures that the cells/particles 108 do not significantly diffuse during the exposure time, thereby blurring the image.

For impulse-response measurements, appropriate impulse generating apparatuses 117 may be attached to the microscope 106 in order to apply forces to the cells/particles 108 that are synchronized with image acquisition in order to measure their responses.

FIG. 33A shows bead intensity fluctuations (NSD, normalized standard deviation values) of two different samples of bead (4.8 μm silica) dispersions measured without the application of any external forces (i.e., thermal equilibrium measurement), where one sample was bound to the surface (dense cross-hatched histogram) and the other was unbound to the surface (histogram with sparse diagonal line pattern). The beads were made to bind to the surface of a glass coverslip by diluting them in a 0.9% saline solution, while the beads were made to freely to diffuse on the coverslip by dilution in deionized water. The unbound beads are distinct from the bound ones under these experimental conditions (i.e., minimal overlap of distributions).

FIG. 33B shows the same samples measured after the application of stage movements. Each image in the stack of analyzed histograms was taken after a piezo-electric stage moved the sample 25 μm back and forth along one axis. This physical perturbation on the system did not affect the mobility of the bound population of beads (dense cross-hatched histogram) in comparison to the measurements taken without force application, as measured by the NSD measurements. However the unbound beads displayed higher NSD values than their non-forced counterparts in FIG. 33A as a result of the increased bead mobility generated by the stage motions. Physical force application can thus better resolve bound and unbound populations as evidenced by the greater separation between the bound and the unbound bead histograms in FIG. 33B. Although the resolution between bound and unbound bead populations was excellent even without force application (i.e., thermal equilibrium measurement FIG. 33A), there may be samples and circumstances which warrant the greater resolving capability of bead or cell mobility measurements under the influence of physical force.

All aspects of the testing, from robotic sample container manipulation, to pump or micropipette delivery of samples and their temperature control, to focusing and sampling of different regions of the sample 107, to synchronized force application, image acquisition and data analysis may be automated and computer-controlled (i.e., by computer 113), yielding measurements and results without human intervention.

Thermal Equilibrium Measurement Method

In another exemplary embodiment, analysis of the statistical properties of the sequence of holographic images in the passive probe method, yields a measurement that is related to the extent of the cell/particle 108 motion generated at thermal equilibrium in the sample 107. In this embodiment, a passive probe method is utilized, where the sample 107 is measured by analyzing the fluctuations in microscopic cell/particle 108 motions at thermal equilibrium (e.g., equilibrium fluctuation measurement), using a modified apparatus of FIG. 1A. In lieu of a separate, externally applied physical probe technique, the passive probe method, conducted at thermal equilibrium, uses the thermal energy of the fluid molecules surrounding the beads/cells 108, causing the free beads/cells 108 to undergo Brownian motion. The extent of this Brownian motion may then be measured with the techniques described herein to draw conclusions about the sample.

In the passive probe method, as in the physical force application method stated above, the measurement methods of this embodiment involve acquiring sequences of holographic images of cells/particles 108 in a field of view (e.g., as viewed through a microscope 106) (see FIG. 2).

Specifically, in the thermal equilibrium method, a microfluidic (or microtiter plate) based apparatus 116 is used, where a flow of particles/cells 108 enters into a sample chamber 118, where a sufficiently dilute dispersion of the particles/cells 108, which has settled to the bottom, optically transparent imaging surface 119 of the sample chamber 118, is illuminated with the laser light source 100. The microfluidic (or microtiter plate) apparatus 116 may allow for multiple solutions, and dispersions to be mixed, incubated and measured in appropriate chambers 118. The microfluidic (micropipettiter plate) apparatus 116 will also contain necessary pumping (pipetting) capabilities, temperature control capabilities, and possibly centrifugation capabilities, known to one of ordinary skill in the art.

FIG. 4, screen shot A, shows a holographic image of 2 μm silica beads. A sequence of such holographic images are acquired by the imaging device 112 and statistically analyzed by the computer 113.

FIG. 5A illustrates how each pixel position through the stack is analyzed to determine its standard deviation as well as its average pixel value. A holographic fluctuation image is then generated by the computer 113, whereby each pixel value is given by the value of the pixel's standard deviation over time, divided by the pixel's average value.

In one embodiment, a statistical image(s) is generated by some combination of pixel statistical measures including average pixel value, pixel standard deviation, pixel variance, higher order pixel fluctuations, pixel temporal correlation functions, pixel spatial correlation functions, pixel spatio-temporal correlation functions, background pixel value, background pixel standard deviation, background pixel variance, higher order background pixel fluctuations, background pixel temporal correlation functions, background pixel spatial correlation functions, and background pixel spatio-temporal correlation functions. The pixel statistical measures may be generated pixel-wise over the image sequence and then averaged over the particle neighborhood. The pixel statistical measures may be calculated over the particle neighborhood, and then calculated over the corresponding neighborhoods in the other images in the sequence. The particle neighborhood mask may be generated using one frame in the sequence. Multiple frames from the sequence may be used to generate particle neighborhood masks. Pixel statistical measures may be calculated over subsets of the image sequence. Pixel statistical measures may be calculated over successive subsets of the image sequence generating time varying statistical measurements per particle. Time varying statistical measurements per particle may be associated with time varying experimental conditions (e.g., physical movement, vibration, solution conditions, flow conditions, other environmental effects). Spatial and temporal correlations of particle-based statistical measurements may be performed. Time varying spatial and temporal correlations of particle-based statistical measurements may be calculated (as opposed to pixel-based statistical measurements). Thresholds in the statistical particle measurements may be chosen to select fractions with desired surface affinity, interaction characteristics that have commercial, diagnostic relevance. Thresholds in selected fractions may be chosen to indicate minimum fractional level necessary to be measured before positive result is indicated, based on desired statistical significance.

In one embodiment, particle positions are tracked over time, and statistical measurements of particle positions may be generated (e.g., mean squared displacement, net displacement, etc.) and distributions of these quantities plotted for multiple particles. Thresholds in particle position measurements and particle statistical quantities based on particle positions may be applied to determine fractions with target particle-surface affinities. Statistical measures of particle movement may be based on mean squared particle displacement, mean particle displacement, net particle displacement, higher-order particle position statistics, or a combination of any or all of these quantities. Similar measures of particle movement may be measured for control purposes (e.g., background correction).

An example of a holographic fluctuation image generated from a sequence or stack of holographic images (i.e., 60 images), as discussed above with respect to FIG. 5A, is shown in FIG. 4, screen shot B. The light areas in the image correspond to areas in the sample plane where there are large fluctuations in signal intensity (corresponding to where particles/cells 108 are moving), while the dark background demonstrates that the fluctuations in regions where there are no particles/cells 108 present, are much lower.

Accordingly, FIG. 5A shows an analysis of a stack of holographic images, yielding the holographic fluctuation image S(i, j) which is constructed by dividing the pixel-wise standard deviation by the pixel-wise average, for each pixel, over the stack of images.

More specifically, the fluctuation image, S(i,j), is a representation of the normalized standard deviation distribution (i.e., standard deviation divided by the mean, pixel by pixel) for the stack of images. This image is then processed to generate a distribution of average normalized standard deviations over each cell/particle 108. It is this distribution, or histogram of normalized standard deviations of cells/particles (NSD) 108 which yields the desired information about the sample 107 mobility and hence the particle-surface interactions.

In another embodiment, which utilizes the apparatus of FIG. 1A as shown, 2 μm silica beads 108 are dispersed in water, diffusing on the coverslip surface 109, and holographic fluctuation images (i.e., images of normalized standard deviations over the sequences of images) are shown in FIG. 6, screenshot A and B. Data was taken on a temperature-controlled microscope stage 114.

Specifically, FIG. 6, screenshot A, is a holographic fluctuation image (i.e. normalized standard deviation image) of the sample at 5° C. (constructed from a 60 frame stack), and FIG. 6, screenshot B, shows the sample's 107 holographic fluctuation image of 2 μm silica beads in deionized water when equilibrated to 47° C. (constructed from a 60 frame stack). Each holographic fluctuation image is generated from 60 images acquired over approximately 11 seconds.

It can be seen from the screenshots that the regions of high intensity at the lower temperature (FIG. 6A) are more confined compared to the high intensity regions at the higher temperature (FIG. 6B). This indicates that the particle motion generating the high intensity fluctuations is more confined at the lower temperature. This is expected from diffusion theory since the mean-squared displacement of a particle is decreased as the temperature decreases.

FIG. 7 shows three histograms of normalized standard deviations of 4.8 μm diameter silica beads in water on a glass coverslip (40 frames acquired at 5 frames per second), measured at temperatures of 6.8° C. (lower histogram), 28° C. (middle histogram), and 46.3° C. (top histogram). The distribution measured at the lowest temperature (6.8° C.) also has the lowest values of the normalized standard deviation (NSD). The main peaks in the histograms correspond to the freely diffusing beads 108, while the minor peak at lower values of NSD (5-10%) represent the much smaller fraction of beads 108 that are stuck or partially stuck to the surface.

The main peak shifts to higher values of NSD as the temperature is increased. The beads at higher temperatures undergo higher-amplitude thermally induced motion, causing greater pixel intensity fluctuations which are detected as higher bead NSD values.

Specifically, for free diffusion, the mean squared displacement is linearly proportional to the temperature:

Δx²=4Dt, D=k_BT/(6πηr),

where D is the diffusion coefficient (or effective diffusion coefficient), t is the time interval between particle position measurements, k_B is Boltzmann's constant, T is the temperature in Kelvin, η is the viscosity (or effective viscosity) and r is the radius of the spherical diffuser. The viscosity of water decreases as the temperature increases, causing an additional temperature dependence that increases the mean squared displacement as the temperature increases. FIG. 8 is a plot of the mean value and standard deviation of the normalized standard deviation distributions of the 4.8 μm silica beads freely diffusing on a glass coverslip at different temperatures (open squares). The upper x-axis reflects the temperature the measurements were taken at, while the lower x-axis shows the theoretically expected square-root of the mean squared displacement, given the bead diameter, temperature (which was placed on a temperature controlled sample stage), temperature dependent viscosity of water, surface-viscosity correction (Faxen's Law) and time interval. The measured NSD value is plotted against the temperature (upper axis) as well as the corresponding square-rooted mean squared displacement (lower axis) calculated based on the diffusion formula (see above) and a surface viscosity correction factor (Faxen's Law).

The second plot in FIG. 8 (black squares), shows the mean and standard deviation of simulated bead motion and generated by displacing sub-images of beads by step-sizes that are governed by a Gaussian distribution whose mean is zero and that has various different widths (corresponding to various square-rooted mean squared displacements). Forty such displaced sub-images are generated, simulating a single bead's image stack. An ensemble of 250 such simulations was performed for a range of step-sizes (i.e., square-root mean square displacements). The identical algorithm for calculating the beads normalized intensity fluctuations (normalized standard deviation (NSD)) that was applied to the experimental data (which was 40 frames in duration) was then applied to the simulated data, generating a distribution of simulated NSD values at each step size, the mean and standard deviation of which are plotted.

The comparison between the measured and simulated values is excellent, confirming the choice of the normalized standard deviation as an excellent measure of quantifiably measuring the changes in the mean-squared displacements of the beads as a function of temperature, and as an alternative to applying particle tracking algorithms. As noted above, the slightly larger error bars and lower average value of the normalized standard deviation values for the measured beads may be explained by the small fraction of bound beads in the largely unbound population (note small peak between 5-10% NSD in histograms of FIG. 7).

There are a number of advantages to using the normalized standard deviation as a measure for quantifying particle mobility. First, only one image, representing a spatial map of the pixel-wise statistics through the stack, is required to reflect the dynamics of the collection of particles 108. This statistical image related to pixel fluctuations may also be generated in real-time, as the computational load required to compute it is relatively low. Furthermore, calculating this statistical image does not require any assumptions concerning cell/particle shape.

Second, particle positions may be established using only one frame from the sequence. To assign fluctuation observables to each particle 108, one additional image from the sequence is required in order to determine the location of each particle 108. The particle neighborhoods are then masked with the statistical pixel fluctuation image to generate average fluctuations per particle for all the particles in the field of view (see FIG. 5B). Since the particles diffuse/move only short distances compared to the average inter-particle distances, one frame is sufficient to establish particle neighborhoods for generating normalized standard deviation distributions from the statistical pixel fluctuation image. This approach allows the computational requirements for object recognition normally applicable to each image in the sequence to be significantly relaxed.

Third, the normalized standard deviation measurement method readily scales to lower magnification allowing more particles to be measured per field of view. This is especially useful when particle binding probability is expected to be low, since it allows more particles to be measured at one time, improving the binding detection statistics of the measurement.

Using a coherent source 100 to illuminate the sample 107 of particles 108 provides additional advantages for sample 107 measurement. While diffusing objects with low contrast under conventional illumination, may present challenges to robust detection and position tracking due to the greater effect of background image noise on the measurement, using a coherent source 100 to image, higher-contrast images may be generated by tuning the amount of defocus. In addition, with a coherent source 100, the focal plane position may be determined numerically (e.g. holographic focusing), obviating the time consuming mechanical focal scan methods traditionally used. This is possible since de-focused samples 107 illuminated by a coherent source 100 generate diffraction patterns which may then be numerically propagated over a range of distances to find the required focus position.

Surface Binding Detection Measurements

In another embodiment consistent with the present invention, an application of the above instrumentation and data analysis technique is the detection of particles 108 binding to a surface 109, as illustrated in FIG. 9. The apparatus of FIG. 1A is used in this exemplary embodiment as well.

For example, FIG. 9 shows a diagram of components of a sample 107 on a treated surface 109, of coated beads 108 with either one type of antigen (A-∘) or another type of antigen (B-□), with the goal of the assay being the selective detection of each type of bead based on its interaction with an antibody coated surface.

FIG. 9, slide 1, shows each type of particle 108 is placed on an antibody coated surface 109. The particle 108 coated with A antigen is specifically bound to the immobilized specific antibodies (A antibodies) coated on the coverslip surface 109, thereby restricting the motion of the particle 108. FIG. 9, slide 2, shows the differing behavior of each type of particle 108, where the B-antigen coated particle 108 which is not specifically bound, is allowed to freely diffuse in the solution, on the surface 109, while the A type particle motion is severely restricted due to specific binding with the surface antibodies.

A similar type of result may be achieved with cells 108. The following discusses experiments conducted on red blood cells, to distinguish their surface antigens based on interactions with specific antibodies coated on a coverslip surface 109, using an apparatus

• as described in FIGS. 1A, 1B, and 3.

For example, red blood cells may be characterized by their ABO blood group. A person may be tested as a Type A, meaning that they have red blood cells (RBCs) with type A antigen on their surface; or as a Type B with type B antigens on their surface; or as a type O having neither antigen present; or as a type AB and have both antigens present on the surface.

In addition to the A and B antigens, red blood cells are also tested for the presence or absence of RhD antigen, with the presence of the antigen denoted by a “+” sign or “positive”, and the absence of the antigen indicated by a “−” sign or “negative”. Tests were performed by imaging the interaction of the patient's red blood cells on three separate regions on the surface 109 using the apparatus of FIG. 1A. The surface 109 included a chemically treated coverslip 109 allowing specific antibodies to be linked to the surface 109. The surface 109 included three regions, one region which had anti-A antibodies linked to the surface 109, one with anti-B linked, and one with anti-RhD linked.

The patient's blood was diluted and then introduced into the sample chamber 118 whose bottom surface 119 was the treated coverslip 109 glass with immobilized antibody patches. Red blood cells settled to the bottom surface 119 and interacted with the particular antibodies present on that patch. Cells 108 with the specific antigen on the surface 119 corresponding to the antibodies present on the underlying patch were specifically bound to the surface 119. Cells 108 without specific antigens corresponding to the underlying immobilized antibody were free to diffuse on the surface 119. This type of testing (i.e., typing of the antigens on the red blood cell membrane surface) is similar to a forward typing of the blood sample (as opposed to reverse typing, which is a complementary typing technique that measures the presence of antibodies specific to surface antigens, in the blood plasma).

The holographic fluctuation method outlined above is used to determine if, a) the red blood cells are specifically stuck to a particular patch, indicating a positive reaction to the antibodies present on that patch, or b) they are freely diffusing, indicating a negative reaction, and therefore a lack of specific interaction. This determination is made based on the measured histograms of cellular normalized standard deviations of their holographic pixel fluctuations. Specifically bound cells 108 in the field of view are reflected by narrow distributions in the normalized standard deviation measure with relatively low average value. Free cells demonstrate a wider distribution with a significantly higher average value of NSD (normalized standard deviation).

FIG. 10A shows plots of normalized standard deviation histograms of two different samples 107 measured on an anti-A coated surface. Red blood cells from a donor that is positive for A antigen on a surface with immobilized anti-A antibody, is indicated in black (A type), while RBCs from another donor that is negative for A antigen is indicated by a hatched pattern (B type donor—i.e., A antigen is not present on cells), on a similarly anti-A coated surface. The type A cells (black) are specifically stuck to the surface, drastically restricting their range of motion as reflected by the low values of the cellular normalized standard deviation. The type B cells (hatched pattern) on the other hand do not bind to the surface, as evidenced by large fluctuations in the cellular pixel values, and the high normalized standard deviation values.

The type B blood cells are not specifically bound to the anti-A surface as demonstrated by the large magnitude and broad distribution of the normalized standard deviations values. The type A blood is specifically bound, as is reflected by the narrow distribution and low magnitude of the normalized standard distribution values.

In addition to clearly differentiating between bound and unbound cells, intermediate levels of binding may also be detectable. FIG. 10B shows plots of normalized standard deviation histograms of two different samples 107 measured on an anti-A coated surface, measured over one hour after measurements in FIG. 10A.

The type A distribution (A positive) in FIG. 10B looks similar to that of FIG. 10A, however the type B distribution (A negative) in FIG. 10B features a second peak, indicated by an increased fraction of cells with intermediate normalized standard deviation values, indicating a partial or intermediate state of binding to the surface, possibly due to a time-dependant non-specific cell-surface interaction.

In FIG. 10B, the Type B donor cells (hatched pattern) are not specifically bound (i.e., negative for antigen A) and have a broad histogram, reflective of diffusive behavior. However a subpopulation with a peak at ˜7% NSD (Normalized Standard Deviation) is apparent after the passage of one hour. These cells may represent a partially bound fraction due to non-specific interactions with the surface that become more dominant with time. Although this subpopulation demonstrates an increased immobilization compared to the other unbound cells, the subpopulation is still distinct from the specifically bound population (i.e., population positive for antigen A), allowing the technique to differentiate specific binding from non-specific binding.

It is apparent that cells with “intermediate” levels of binding are still distinguishable from the specifically bound cells, recommending that this technique can resolve sub-populations of cells interacting differently with the substrate.

Similar data may be collected with the donor cells allowed to interact with patches coated with anti-B as well as anti-RhD immobilized to the surface. In this way each donor's red blood cells may be probed for the presence of A, B, and/or RhD antigens, yielding a forward blood typing test result.

FIG. 11, slide 1, shows two types of particles, a type A particle coated with type A antigen, as well as a type B particle coated with a type B antigen. The surface in this configuration is coated with antigen A. Binding to the surface is possible for type A particles in the presence of A-antibody (e.g. IgM) which may simultaneously bind to both the A antigen coated particle and the A antigen coated surface, thereby immobilizing the particle. The test involves mixing the patient's blood plasma (containing the antibodies), and incubating it with particles with known surface antigens, on surfaces with known antigen type. The goal of the test is to determine which type of antibody is present in the plasma, by measuring the behavior of each type of control particle on each type of surface.

The type of antibody present in the solution may be determined by measuring the mobility of particles on surfaces which are coated with appropriate antigens. Conversely with known antibodies, the type of particles may be determined by measuring their mobility on surfaces coated with the appropriate antigen(s). The A antibody in solution is able to bind to the A antigen coated surface, as well as simultaneously bind to a particle coated with A antigen. FIG. 11, slide 2, shows the differing behavior of each type of particle, where the B-antigen coated particle which is not specifically bound, is allowed to freely diffuse in the solution, on the surface, while the A type particle motion is severely restricted due to specific binding with the antibody that is itself specifically bound to the surface. Thus, FIG. 11, slide 2 shows how the unbound type B particle may be distinguished from the bound type A particle by measuring particle mobility.

FIG. 12A shows a series of NSD histograms for a sample of red blood cells taken at different times. The sample consists of type A red blood cells in the presence of a high concentration of anti-A IgM antibody (100 nM) dispersed in synthetic plasma onto a surface with type B antigens on a glass coverslip surface. This measurement configuration is similar to the reverse blood typing method wherein the plasma of the subject is tested for the presence or absence of naturally occurring antibodies by being able to detect the extent of binding between blood cells of known type on surfaces of known type. The presence of binding in such a scenario indicates the presence of antibodies that are able to simultaneously bind the cells to the surface, thereby immobilizing them. Essentially, forward typing involves detecting the presence of particular antigens on the red blood cell surface, while reverse typing detects the presence of particular antibodies in the blood plasma. Assays may be designed to perform reverse blood typing results based on the detected presence or absence of blood group antibodies which would be reflected in the degree of fluctuations in the bead/cell hologram sequence.

The bottom histogram in FIG. 12A, was measured 13 minutes after type A cells dispersed in synthetic plasma with 100 nM anti-A was introduced onto a surface with B antigens. Each histogram was generated by an analysis (cellular NSD calculation) of a sequence of 40 frames acquired at a rate of 5 frames per second. The cells are able to diffuse, as evidenced by the high average values of the NSD observable. The cells show similar diffusive behavior at later times as well, as may be seen by the similarity of the histograms measured at the 21, 29, 37 and 45 minute time marks.

FIG. 12B shows a time series of histograms under similar conditions to those in FIG. 12A, except that the surface had type A antigens on it (unlike the type B antigen surface measurements of FIG. 12A). This high concentration of anti-A (100 nM) was chosen to ensure that A cells were bound to the A type surface (as was visually confirmed). The first measured histogram (bottom histogram), taken 14 minutes after the cells were introduced onto the surface, show that the cells are immobilized (low average NSD and narrow width), as expected, given the high concentration of antibody in the synthetic plasma. At later times the cells remain bound as well, as seen by the similar NSD histograms at the 22, 30, 38 and 46 minute marks. Comparison of the histograms of FIGS. 12A and 12B allows us to choose a threshold value for NSD to determine whether a cell is bound or not. A threshold value of 7% NSD under these conditions is appropriate, with a majority of the cells in the unbound population of FIG. 12A being above this threshold, and the majority of cells in the bound population of FIG. 12B being below this threshold.

FIGS. 13A and 13B are similar to the conditions measured in FIGS. 12A and 12B except that a much lower concentration of antibody was used, only 1 nM of anti-A in the synthetic plasma was present. In FIG. 13A, the A type cells in synthetic plasma containing 1 nM anti-A dispersed on the B type surface are mostly unbound when first measured at 19 minutes after the sample was introduced onto the surface. Subsequent measurements at the 25, 31, 37, 43, 49, and 55 minute mark show similar behavior. In FIG. 13B the A type cells in synthetic plasma containing 1 nM anti-A were measured after being introduced onto a surface with A type antigens. After 18 minutes, a slightly higher fraction of cells with low NSD values can be detected over the control set of cells (i.e. unbound cells) measured in FIG. 13A. With time, this low NSD fraction increases in magnitude, as can be seen by the progressive leftward shift in the histograms. The low 1 nM concentration of anti-A decreases the rate at which bonds between the cells and the surface are made, in comparison to the much faster binding interaction between the cells and the surface when 100 nM anti-A was used (FIG. 12B), where practically all cells were found to be bound within 14 minutes.

In another embodiment, a succession of binding assays may be performed on given samples, yielding particle adhesion kinetics curves. Such assays may yield information on not only the degree of binding at a given time, but also information on the rate of change of binding, allowing binding kinetics to be modeled. FIG. 14 shows the cell adhesion kinetics for the same sample studied in FIGS. 13A and 13B. At each time point (sample prepared at t=0) the fraction of cells with an NSD value less than the threshold value 7% NSD was calculated for each surface, to estimate the fraction of cells bound on each surface. For the A cells on the B surface, negligible binding is detectable (i.e., ˜0.5%), although it does rise slowly over time (i.e., there exists time-dependent non-specific cell-surface interactions which increases the bound fraction over time). The A cells on the A surface on the other hand show significantly higher binding at the first measured time point, and the bound fraction increases significantly after that over the course of about 35 minutes (i.e., from ˜1% to ˜5%).

By measuring the mobility in a cell by cell fashion, the fraction of cells that are bound may be measured. In contrast to many bulk measurements where a significant fraction of the cells must react to yield a signal, this technique, due to its single-cell sensitivity is able to measure positive reactions when only a fraction of the cells are bound.

Diffusion Modeling (Calibration) and Simulation Results

A connection exists between the particle's normalized standard deviation measurement and its underlying particle position measurements. Particle motion was simulated in order to model experimentally measured NSD distributions.

FIG. 15, slide A, shows an image of a 4.8 μm diameter silica bead. A sequence of 40 frames of this diffusing bead (acquired at 5 frames per second) was analyzed using a particle tracking algorithm which yielded a sequence of shifts in the measured centroid position, Δx and Δy, plotted in FIG. 15, plot B. The standard deviation in the x and y directions was measured to be 0.17 μm (comparable to the theoretically expected value of 0.16 μm calculated from diffusion theory for an equivalently sized bead in water, and including the surface-viscosity correction factor from Faxen's law).

The identical sequence of frames analyzed by particle tracking in FIG. 15 may be reanalyzed by an analysis of the pixel fluctuation using the normalized standard deviation method.

FIG. 16, slide A, shows the image of the NSD calculated from the same measured sequence of 40 frames that was used in the particle tracking analysis shown in FIG. 15B, which yielded a NSD value, averaged over the bead area, of 23.6% NSD.

Simulations using the image of the bead from one frame in the sequence were performed by generating a sequence of 40 displaced bead images, with the displacements governed by a Gaussian distribution whose width was experimentally measured using the particle tracking result from FIG. 15, plot B. An ensemble of 250 such simulated bead sequences was generated, along with 250 calculated NSD images (i.e., one NSD image for each sequence). The averaged NSD value over the bead area for each member of the ensemble is plotted in FIG. 16, plot B, with a mean value of 23.7% NSD and a standard deviation of 2.1% NSD, showing excellent agreement with the experimentally measured NSD (see FIG. 16, slide A).

A series of simulations may be performed to relate the NSD results to particle motions with varying step-sizes (i.e., Gaussian distributions with varying widths).

Such a calibration curve relating the mean of the measured bead NSD distribution to the root mean squared displacement for a series of simulations is shown in FIG. 17 (filled squares). The curve was generated by simulating frame to frame bead shifts with Gaussian distributions that have a range of widths, corresponding to bead motion with a range of mean-squared displacements. Each point in the curve was generated with an ensemble of 250 simulated bead sequences (40 frames each), generating an NSD distribution whose mean (filled squares) and standard deviation (open squares) are plotted. Note that the curve is linear for root mean squared displacements (labeled as “std dev” in the FIG. 17) less than 0.1 μm. For larger displacements, however (0.3 μm-0.4 μm), the NSD values level off and start to decrease in magnitude. This effect is due to the bead going beyond the borders of its neighborhood over the course of 40 frames, thus, causing the fluctuations that are registered in the neighborhood to decrease as well.

FIG. 18 shows the calibration curve plotted in FIG. 17 (filled squares), as well as two other calibration curves generated from two other images of 4.8 μm diameter silica beads (circles and triangles). Although each image is displaced using an identical range of distributions, the curves are significantly different, indicating the difference in intensity distribution of the bead images is the source of the difference. The open symbols indicate the experimentally measured mean NSD values for a 40 frame sequence of each bead. The experimental results track the simulated ones.

FIG. 19 shows the beads from FIG. 18, along with a quantity termed the contrast correction factor. This factor is determined by calculating the standard deviation of the pixel intensities in the neighborhood of each bead (i.e., within the rectangle surrounding each bead), divided by the mean value of the pixel intensities in the same region (i.e., normalized spatial standard deviation). Higher contrast images tend to have higher contrast correction factors. A diffusing bead with a higher contrast correction factor also generates greater pixel fluctuations than a similarly diffusing bead with a lower contrast correction factor, causing a comparatively higher NSD value to be measured. To correct for this image artifact each bead's curve was scaled with respect to the contrast correction factor of bead 1. The rescaling causes the three curves in FIG. 18 to collapse to the curve of bead 1, approximately. The open symbols indicate the rescaled experimentally measured mean NSD values, similarly rescaled. These show a tighter distribution as well. Since the NSD parameter is intended to reflect the dynamics of the beads/cells and not their contrast, factoring out the contrast dependent contribution to pixel fluctuation magnitude (i.e., contrast correction factor) improves the utility of normalized standard deviation (NSD) as a dynamical observable.

Contrast correction may also be applied to a population of cells dispersed on treated surfaces. FIG. 20, plot A, is an NSD histogram of B type red blood cells in synthetic plasma which includes anti-B, dispersed on a surface with A antigens. Since the anti-B antibodies which attach to the B cells do not bind to the A antigens on the surface, they are free to diffuse. FIG. 20, plot B, is a scatter plot that relates the NSD value of each red blood cell measured in FIG. 20, plot A, to its normalized spatial standard deviation (i.e., contrast correction factor). The positive correlation, shown by the fit, indicates that low NSD values may be generated by low normalized spatial standard deviation, irrespective of the dynamics. Since the dynamics of cells/beads is basically independent of the normalized spatial standard deviation, the dynamical observable (NSD) is corrected against dependence on the normalized spatial standard deviation using the linear fit to generate contrast corrected data.

FIG. 21, plot A, shows an NSD histogram of B-type red blood cells in synthetic plasma which includes anti-B, dispersed on a surface with B antigens. The anti-B antibodies attach to the B antigens on the red blood cells, as well as the B antigens on the surface, thereby immobilizing the cells. As a consequence the intensity fluctuations generated are significantly reduced compared to the cells on the A antigen patch (FIG. 20, plot A), as is shown by the lower NSD measurements in FIG. 21, plot A. FIG. 21, plot B is a scatter plot that relates the NSD value of each red blood cell measured in FIG. 21, plot A, to its normalized spatial standard deviation. As in FIG. 20, plot B above, there is a positive correlation. Using the linear fit, the NSD values are corrected against their spurious dependence on normalized spatial standard deviation values.

FIG. 22, plots A and B, show corrected and uncorrected histograms of unbound (FIG. 20, plot A) and bound (FIG. 21, plot A) red blood cell NSD histograms, where corrections were applied by using the linear fits (FIG. 20, plot B, and FIG. 21, plot B) to remove the NSDs dependence on the normalized spatial standard deviation. The correction does not significantly alter the overall shapes of the histograms. The most significant effect this has is on attenuating the trailing tail of the unbound NSD distribution. Thresholds may be applied to NSD values in order to determine whether cells/beads are “bound” (i.e., “bound” fraction=fraction of cells in NSD histogram<threshold; “unbound” fraction=fraction of cells in NSD histogram>threshold). The attenuation of the trailing edge for the corrected NSD distribution implies that there will be a smaller fraction of cells that are “bound” when threshold values in the tail region are used, thereby decreasing the so called “non-specifically bound” fraction of the unbound population. This feature of the correction is equivalent to a lowering of the “non-specific background” of the test. Another option for minimizing artifacts due to contrast contributions to the NSD is to set up minimum and/or maximum thresholds on allowable spatial standard deviations, and only analyzing those particles within the threshold(s).

FIG. 23 is a plot of the cumulative probability distribution of the histograms shown in FIG. 22, plot A, and FIG. 22, plot B (i.e., a plot of the fraction of cells that have an NSD less than or equal to that particular NSD value). The corrected “unbound” histogram has a significantly smaller fraction of cells with NSD values less than or equal to a threshold value of 9 (0.93%), compared to the uncorrected fraction of 1.7%. This has the effect of lowering the background value against which red blood cell binding quantities may be compared.

Heterogeneous Diffusion Dynamics Detection and Characterization

Uniformly sized beads diffusing on a uniform surface should display diffusion expressible by a single diffusion coefficient. FIG. 24 shows an NSD histogram of 4.8 μm diameter beads in water on a glass coverslip (top histogram) as well as a simulated NSD histogram using a Gaussian distribution that yields a square-root mean squared displacement of 0.131 μm (bottom histogram). The simulated histogram compares well with the free beads in the measured histogram (i.e., NSD>7%). The small fraction of beads that are measured to be stuck (due to non-specific binding) are not included in the comparison (i.e., NSD<7%).

By increasing the salt concentration to 0.4% saline, the beads may be made to stick to the surface. FIG. 25 shows an NSD histogram of 4.8 μm diameter beads in water with 0.4% saline (top histogram) as well as a simulated NSD histogram using a Gaussian distribution that yields a square-root mean squared displacement of 0.0064 μm (bottom histogram). The simulated histogram compares well to the measured histogram for these beads which show much greater restriction in their motion. That is, beads that appear to be bound are in fact undergoing diffusion-like motion with much smaller amplitude, and may be modeled as such. Thus, bound beads may be characterized by their degree of binding, such that beads with smaller root-mean squared displacements (and consequently smaller NSD values) are more highly bound.

FIG. 26, plots A, B, and C, show an NSD histogram of type A red blood cells in synthetic plasma containing 5 nM anti-A dispersed on a surface with B antigen (FIG. 26, plot A, top histogram, 2653 cells measured). The anti-A antibody cannot make bonds bridging the cells to the surface, allowing the cells to diffuse. A simulation of RBC diffusion with a single characteristic root mean-squared displacement was found to model the average NSD value, when a root mean-squared displacement of 0.069 μm was used for the width of the Gaussian step-size distribution. The simulation was performed 2650 times to get a distribution of NSD values (FIG. 26, plot B, middle histogram). Note that the measured histogram is significantly broader than the simulated histogram, indicating that a single root-mean displacement does not characterize the dynamics of the RBCs on the surface. FIG. 26, plot C, shows a simulated NSD distribution (2650 trials) using a Gaussian distribution of root mean-squared displacement values (0.069±0.02 μm), modeling the measured cellular diffusion as a heterogeneous distribution of cell diffusivities. Modeling the diffusivities as a distribution of freely diffusing spheres that yield an identical distribution of root mean-squared displacements, yields a Gaussian distribution of radii that vary by more than a factor of three (within one standard deviation of the mean radius), implying significant heterogeneity in the diffusion dynamics. Since the size distribution of the cells is much more narrow, this result indicates heterogeneous interactions of the cells with the surface, due to surface and/or cellular heterogeneities that cause heterogeneities in the interaction potentials of the cells on the surface (i.e., diffusion is not free, and it is not uniform). Thus, the invention allows the detection of heterogeneous as well as homogeneous diffusion dynamics, which may be used as a diagnostic of the distribution of cellular properties in the presence of known surface conditions. In particular, such measurements and analysis may yield information of the heterogeneity of cell (particle) properties (e.g., surface binding density, affinity etc.) for research or diagnostic purposes, in the presence of known surfaces. For example, a population of cells may express a range of surface proteins with a range of expression levels. Determining the fraction of cells expressing a given density of surface proteins may be of diagnostic interest. Cells expressing a higher surface protein concentration will be more strongly bound to a complementary surface than cells with a lower surface protein concentration, as reflected in their mobility observable (e.g. cellular normalized standard deviation value). Furthermore, having detected the target fraction of cells, they may be isolated and/or tested with other surfaces and under other conditions for generating information of additional diagnostic value. Such cells may also be isolated for further manipulation (e.g., genetic) as well as culturing purposes. Culturing of selected cells may be for the purposes of expressing and purifying therapeutically useful cellular components (e.g. proteins).

Complementarily, such measurements and analysis may yield information of the heterogeneity of surface properties (e.g., for quality control, inspection etc.) when measured in the presence of known particles. For example, by using calibration particles with a given surface density of antigens on test surfaces, a homogeneous mobility response from the population indicates a homogenous surface, while a heterogeneous mobility response indicates a surface with some degree of heterogeneity (e.g. non-uniform surface density of surface linked moieties).

Cellular Spatial Statistics and Cell Shape Change Dynamics

Information about the state of binding may also be found in the spatial normalized standard deviation of the red blood cells. FIG. 27, plot A and plot B, are plots of the normalized spatial standard deviation values for the unbound and bound red blood cells, measured in FIG. 20, plot A, and FIG. 21, plot A, respectively. The bound population of red blood cells has a slightly lower mean value, though there is significant overlap with the unbound population's normalized spatial standard deviation values. Cells that are bound to the surface are slightly flattened compared to the unbound cells, which feature a biconcave geometry typical of red blood cells. This slight flattening in turn decreases the contrast of the cell by a small amount. Although there is an average difference in the populations, the distributions of the bound and unbound spatial normalized standard deviation values overlap significantly. The normalized spatial standard deviation of each particular cell does not vary significantly from frame to frame over the duration of the measurement (see FIG. 28, plot A and FIG. 28, plot B), justifying our calibration method outlined above.

FIG. 28, plot A, and FIG. 28, plot B, plot the average spatial normalized standard deviation for 100 cells in the unbound population of FIG. 20, plot A, and FIG. 24, plot A (FIG. 28, plot A), as well as that for 100 cells in the bound population of FIG. 20, plot B, and FIG. 24, plot B (FIG. 28, plot B) over each of the 40 frames in the sequences. In both of the plots the error bars indicate the standard deviation of each cell's spatial normalized deviation over the 40 frame sequence. The variation of a cell over frames is insignificant compared to the variation over cells.

Information about the cellular shape change dynamics may also be extracted and utilized by an analysis of the cellular spatial intensity statistics over a sequence of images. Cells that are “floppier” (i.e., change their shape more readily due to either thermal fluctuations or externally applied forces) should demonstrate higher spatial intensity fluctuations compared to cells that are relatively more rigid. Note that the intensity fluctuations described in this section (spatial intensity) are calculated in each cell's frame of reference so that center of mass cell motion (i.e., the cellular mobility measured using the previously described normalized standard deviation measurement) does not interfere with the calculation of the cellular spatial dynamics.

FIG. 29 plots the standard deviation of the spatial normalized standard deviation measured for each cell, over the 40 frames in the sequence for both the bound cells as well as the unbound cells. Even though the magnitude of the frame to frame changes in the normalized spatial standard deviations are small, there turn out to be significant differences between the variation in normalized spatial standard deviation of bound and unbound red blood cells. The bound cells have a broader distribution and have significantly more low-amplitude varying cells, compared to the unbound cells, indicating that a higher fraction of bound cells have low amplitude spatial intensity fluctuations, compared to unbound cells. Nineteen percent (19%) of the bound cells in FIG. 29 have a standard deviation of the normalized spatial standard deviation (measured over 40 frames taken at 5 frames per second) less than 0.005, while among the unbound population only 0.3% of the cells have a value less than 0.005. Unlike the analyses done previously, cell neighborhoods are determined in each frame of the 40 frame sequence (as opposed to determining the cell neighborhood in only 1 frame from the sequence) to correct for center of mass motion contributions to fluctuations in the spatial standard deviation. Such distributions of the spatial statistics of cells over a sequence of frames may be used as a diagnostic of membrane flexibility, cytoplasmic elasticity/visco-elasticity, cell health/disease, cell age, solution conditions, binding state, etc., all of which may influence the dynamics of cellular shape changes which are reflected in the dynamics of the cellular spatial statistics over time and in the distributions of such quantities. Frame exposure times must be short enough to freeze the image of the fluctuating cell, if the frame exposure time is too long then cellular motions will be averaged over during the exposure, decreasing the sensitivity of the technique.

Correlated Motion Detection

The above analytical technique is sufficient for cases which do not require the detection of correlation of bead motions. However, assays may be based on detecting bead-bead binding, as opposed to bead-surface binding. Positive binding events (i.e., binding of two or more beads) require a detectable difference between correlated bead motion and uncorrelated bead motion. Beads that are bound to each other will have their movements correlated in time, while beads that are unbound, though neighboring each other, will not demonstrate correlated bead motion.

Accordingly, in another embodiment consistent with the present invention, one method of distinguishing neighboring beads/cells that are bound to each other as opposed to just resting next to each other is as follows.

In this method, for each neighboring pair of cells/particles, the correlated fluctuations are calculated by multiplying (or alternatively, summing) the neighborhood of each pair of cells/particles sub-images frame by frame. This yields a “correlation neighborhood”, which can be quantified by calculating its average normalized standard deviation.

Fluctuations in the correlation neighborhood increase in amplitude when there are in-phase motions of pairs of beads that are bound together.

Just as the mobility of independent beads/cells may be quantified by comparing the extent of the average fluctuations in each bead/cell neighborhood, the correlated mobility of each adjacent bead/cell pair may be quantified by comparing the extent of the average fluctuations for each correlation neighborhood.

The technique for detecting correlated motion involves, in step 200, detecting each bead/cell 18 in each holographic image 120. Beads/cells 18 that are adjacent in the image are candidates for being bound together and thus demonstrating correlated motion. Upon establishing the beads/cells 18 that have adjacent neighbors in step 201, a list of bead/cell pairs may be made in step 202.

Each adjacent bead/cell 18 pair 121 is analyzed by extracting the image of each bead/cell 18 in the pair 121, creating two sub-images, in step 203. The two sub-images are then multiplied together in step 204, yielding a product sub-image. This procedure is repeated for all pairs of adjacent cells/beads 18 in step 205, and then for all frames 122 in the sequence as well in step 206, resulting in a sequence of product sub-images corresponding to each adjacent bead/cell 18 pair 121 (see FIG. 32A). Corresponding to each such sequence for each bead/cell 18 pair 121, the pixel-wise standard deviation of the product sub-image sequence is computed in step 207, which finally yields an average standard-deviation generated by calculating the average value of the pixel-wise standard-deviation divided by the average value for each pixel over the whole product sub-image, in step 208 (see FIG. 32B, the correlation figure).

The average normalized standard-deviation (NSDP, normalized standard deviation of product sub-images) is a single number attributed to a single adjacent cell/bead 18 pair in the sequence of images 120, 122. Adjacent pairs 121 of cells/beads 18 that are bound together will have correlated motions which will tend to increase the corresponding NSDP value, than those that are not bound together, which may thus diffuse independently to yield NSDP values that are lower, due to uncorrelated motion. On this basis, bound cell/bead 18 pairs 121 may be distinguished from unbound cell/bead 18 pairs 121.

Thus, a statistical analysis is used to determine particle binding, which includes processing a holographic fluctuation image to generate an average normalized standard deviation for each particle, which is a measure of particles mobility on given surface which may be a diagnostic of particle properties, solution properties, surface properties or some combination of such properties, for research, industrial, and/or clinical purposes, and which may include an analysis of each particle's spatial statistics over a sequence of frames, which may be a measure of shape change dynamics, and diagnostic of particle properties, solution properties, surface properties or some combination of such properties, for research, industrial, and/or clinical purposes.

Particles not bound to antibodies on said surface demonstrate large magnitude and broad distribution of said normalized standard deviation distributions, and particles which are bound to antibodies on the surface demonstrate narrow distribution and low magnitude of normalized standard deviation distributions.

In one embodiment, the particles are red blood cells, but may also be cells other than red blood cells. Intermediate levels of particle binding are detectable, as indicated by a fraction of particles with intermediate values of normalized standard deviations.

Distributions of normalized standard deviations are analyzed for diffusional heterogeneity to characterize particle properties, solution properties, surface properties or some combination of such properties, for research, industrial, and/or clinical purposes.

It should be emphasized that the above-described embodiments of the invention are merely possible examples of implementations set forth for a clear understanding of the principles of the invention. Variations and modifications may be made to the above-described embodiments of the invention without departing from the spirit and principles of the invention. All such modifications and variations are intended to be included herein within the scope of the invention and protected by the following features.

Claims

1. A method of determining interactions between a plurality of particles and a surface of a sample holder, comprising:

applying a physical force to at least one of a sample of particles on the sample holder, or to the surface of the sample holder, using a physical force application means;

illuminating said particles using an illumination source of an imaging apparatus having a microscope with a field-of-view;

measuring a response of said particles to said physical force application means by acquiring a sequence of images of said particles in said field-of-view using said imaging apparatus, said acquisition of said images being synchronized with said physical force application means;

statistically analyzing, using a processor of a computer system, said images of said particles captured by said imaging apparatus;

wherein said images include a first component that is diffracted by said particles, and a second component that is undiffracted by said particles, and said two components interfere in an imaging plane, yielding an interference pattern produced by said processor, that represents particle by particle intensity fluctuation values; and

wherein said particles that are able to move through one of said physical force application means or diffusion, display high intensity fluctuations, and those that are bound to the surface of the sample holder, display low intensity fluctuations, yielding a nature of the interactions on the surface of the sample holder.

2. A method of performing holographic optical focusing on a plurality of particles in a sample chamber, comprising:

illuminating a sample of particles in a transparent sample chamber using a coherent light source of an imaging apparatus;

acquiring images of said particles using a focusing camera;

displaying images of an out-of-focus diffraction pattern of said particles on a display;

performing numerical focusing of an imaged hologram of one of said images using a processor of a computer system, to determine a focal plane of said particles;

wherein said numerical focusing includes a propagation of said out-of-focus image to different distances which allows a focus measure to be determined numerically by said processor;

associating said focus measure with each numerically propagated image, using said processor, such that an extremum in said focus measure with each numerically propagated image can be found; and

allowing said computer system to perform a single stage movement of said sample chamber to position said sample in a required focal position.

3. A method of determining interactions between a plurality of particles and a surface of a sample holder, comprising:

illuminating a sample of particles disposed on a transparent bottom surface of a fluidic flow device, using an illuminating source of an imaging apparatus having a microscope with a field-of-view;

measuring a movement of said particles at thermal equilibrium by acquiring a stack of images of said particles in said field-of-view using said imaging apparatus;

statistically analyzing, using a processor of a computer system, said images of said particles captured by said imaging apparatus;

wherein said statistical analysis includes determining each pixel position through said stack of images, to determine each pixel's standard deviation and its average pixel value;

generating a fluctuation image of each said pixel, using said processor;

wherein said fluctuation image is a representation of a normalized standard deviation distribution for each said pixel, for said stack of images;

processing said fluctuation image, using said processor, to generate a distribution of average normalized standard deviations over each of said particles;

wherein relatively larger fluctuations in signal intensity indicate said particles are moving, and relatively smaller fluctuations in signal intensity indicate said particles are immobilized by surface interactions; and

yielding information about a mobility of said sample and the interaction of said particles on the surface of the sample holder.

4. A method of determining interactions between a plurality of particles and a surface of a sample holder, comprising:

illuminating a sample of particles disposed on a transparent bottom surface of a fluidic flow device, using an illuminating source of an imaging apparatus having a microscope with a field-of-view;

measuring a movement of said particles at thermal equilibrium by acquiring a stack of images of said particles in said field-of-view using said imaging apparatus;

statistically analyzing, using a processor of a computer system, said images of said particles captured by said imaging apparatus;

wherein said statistical analysis includes determining each pixel position through said stack of images, to determine each pixel's standard deviation and its average pixel value;

generating a fluctuation image of each said pixel, using said processor;

wherein said fluctuation image is a representation of a normalized standard deviation distribution for each said pixel, for said stack of images;

processing said fluctuation image, using said processor, to generate a distribution of average normalized standard deviations over each of said particles;

wherein relatively larger fluctuations in signal intensity indicate said particles are moving, and relatively smaller fluctuations or a lack of fluctuations in signal intensity indicate said particles are immobilized by surface interactions; and thereby

yielding information about a mobility of said sample and the interaction of said particles on the surface of the sample holder.

5. A method of selective detection of different types of particles on a surface of a sample holder, comprising:

introducing a sample of particles in a solution onto an antibody coated surface of the sample holder, said particles being coated with either one or another type of antigen;

wherein particles coated with one type of antigen are specifically bound to immobilized specific antibodies coated on the sample holder, thereby restricting a motion of said particles;

wherein particles coated with another type of antigen are not specifically bound to the sample holder, and said particles freely diffuse in said solution on the surface of the sample holder;

illuminating said sample of particles disposed on the sample holder, using an illuminating source of an imaging apparatus having a microscope with a field-of-view;

measuring a movement of said particles at thermal equilibrium by acquiring a stack of images of said particles in said field-of-view using said imaging apparatus;

statistically analyzing, using a processor of a computer system, said images of said particles captured by said imaging apparatus;

wherein said statistical analysis includes determining each pixel position through said stack of images, to determine each pixel's standard deviation and its average pixel value;

generating a fluctuation image of each said pixel, using said processor;

wherein said fluctuation image is a representation of a normalized standard deviation distribution for each said pixel, for said stack of images;

processing said fluctuation image, using said processor, to generate a distribution of average normalized standard deviations over each of said particles; and

wherein said specifically bound particles in said field-of-view exhibit relatively lower magnitude and relatively narrower distributions in a normalized standard deviation measure with relatively lower average value, and freely diffusing particles exhibit a relatively higher magnitude and relatively wider distribution with a substantially relatively higher average value of said normalized standard deviation distribution; thereby

determining said one or another type of antigens on said particles or said specific antibodies on the sample holder.

6. A method of detection of particle shape as a diagnostic procedure, comprising:

illuminating a sample of particles disposed on a sample holder, using an illuminating source of an imaging apparatus having a microscope with a field-of-view;

measuring a shape of said particles at thermal equilibrium by acquiring a sequence of images of said particles in said field-of-view using said imaging apparatus;

statistically analyzing, using a processor of a computer system, said images of said particles captured by said imaging apparatus;

wherein said statistical analysis includes determining a spatial distribution of intensity over said particles over said sequence of images;

generating a fluctuation image using each said pixel, using said processor;

wherein said fluctuation image is a representation of a normalized standard deviation distribution for each said pixel, for said sequence of images;

processing said fluctuation image, using said processor, to generate a distribution of average normalized standard deviations over each of said particles;

wherein said particles that change their shape more readily due to one of thermal fluctuations or externally applied forces demonstrate relatively higher spatial intensity fluctuations compared to said particles that are relatively more rigid; and

wherein such normalized standard deviation distributions of said statistical analysis of said spatial intensity of said particles over said sequence of images may be used as a diagnostic of at least one of particle flexibility, elasticity/visco-elasticity, health or disease, age, solution conditions, or binding state.

7. A method of determining an interaction between particles in a sample holder, comprising:

(1) illuminating a sample of particles disposed on a sample holder, using an illuminating source of an imaging apparatus having a microscope with a field-of-view;

(2) measuring a shape of said particles at thermal equilibrium, using a processor of a computer system, by acquiring a sequence of images of said particles in said field-of-view using said imaging apparatus;

(3) detecting each of said particles in each of said images using said processor;

(4) establishing that a pair of said particles that are adjacent to one another in each of said images, may be bound together and may demonstrate correlated motion, using said processor;

(5) extracting an image of each of said adjacent pair of particles, using said processor, creating two sub-images;

(6) multiplying said two sub-images together, using said processor, to yield a product sub-image;

(7) repeating steps (3)-(6) for all pairs of adjacent particles in all of said images, resulting in a sequence of product sub-images corresponding to each of said pair of adjacent particles;

(8) calculating, using said processor, a pixel-wise standard deviation of said sequence of product sub-images;

(9) generating an average standard-deviation for one of said pair of adjacent particles by calculating, using said processor, an average value of said pixel-wise standard-deviation divided by an average value for each pixel over a whole product sub-image;

wherein adjacent pairs of particles that are bound together have correlated motions which increase a corresponding normalized standard deviation of product sub-image values, than those adjacent pairs of particles that are not bound together, to yield said normalized standard deviation of product sub-image values that are relatively lower with relatively narrower distribution, than unbound particles which exhibit uncorrelated motion, wherein said normalized standard deviation of product sub-image values are relatively higher with relatively broader distribution, such that bound pairs of particles are distinguished from unbound pairs of particles.

8. An apparatus for measuring, testing and characterizing a population or sub-population of particles based on their detected mobility, comprising:

an imaging forming apparatus, including: a coherent light source which emits a light beam; and a collimator which collimates said light beam from said coherent light source;

a transparent sample holder of a microscope on which a sample is disposed and which is illuminated by said collimated light beam, said sample which comprises a dispersion of particles; and

means for measuring a mobility of said particles on said sample holder, in order to infer a presence or absence of interactions of said particles with said sample holder.