Abstract: A system and method for segmentation of anatomical structures in MRI volumes using graph cuts is disclosed. In this method, a template is registered to an MRI brain volume. The template identifies seed points of anatomical brain structures, such as the cerebrum, the cerebellum, and the brain stem, in the MRI brain volume. Any or all of the anatomical brain structures can be segmented using graph cuts segmentation initialized based on the seed points identified by the template. It is possible to segment each of the anatomical brain structures by performing a hierarchical three-phase segmentation process including brain/non-brain segmentation, cerebrum/cerebellum and brain stem segmentation, and cerebellum/brain stem segmentation.

Abstract: A method and system for segmenting tubular structures in 3D images is disclosed. User inputs identifying a first region on the image inside of a tubular structure and a second region of the image outside of the tubular structure are received. Based on this information, an ordered series of pearls are generated along the tubular structure. Pearls are spheres, each having a center location and a radius determined based on the center locations and radii of previous pearls and on local voxel intensities in the image. A continuous model of the tubular structure can be generated by interpolating the center locations and radii of the ordered series of pearls. The ordered series of pearls can be displayed and easily edited in response to user input, thus providing an efficient and flexible method for interactive segmentation of a potion of interest in a tubular structure.

Abstract: A method and system for segmenting tubular or stroke-like structures in 2D images is disclosed. Examples of such structures include, but are not limited to, blood vessels, bones, roads, rivers, electrical wirings, and brush-strokes. User inputs identifying a first region on the image inside of a tubular structure and a second region of the image outside of the tubular structure are received. Based on this information, an ordered series of pearls are generated along the tubular structure. Pearls are 2D disks, each having a center location and a radius determined based on local pixel intensities in the image. A continuous model of the tubular structure is generated by interpolating the center locations and radii of the ordered series of pearls.

Abstract: A method and system for side detection of an undetailed 3D ear impression is disclosed. In order to determine whether a received 3D undetailed ear impression is a left or right ear impression, a local coordinate system of the 3D undetailed ear is defined based on side independent features of the 3D undetailed ear impression. A skeleton (or center spline) of the 3D undetailed ear impression is detected, and it is determined whether the 3D undetailed ear impression is a left or right ear impression based on the skeleton and the local coordinate system.

Abstract: A method of computing a continuous interpolation of a discrete set of three-dimensional (3D) balls, including generating an initial skin, wherein the initial skin is a surface comprised of splines and wherein the splines touch each ball along a circle that is tangent to the ball, solving a first differential equation to minimize the initial skin's surface area or solving a second differential equation to minimize a squared mean curvature of the initial skin's surface, wherein the result of solving the first or second differential equations is an updated skin; and repeating the steps of solving the first or second differential equations for the updated skin, and then, repeating the steps of solving the first or second differential equations for each subsequently updated skin until a desired skin is realized.

Abstract: A system and method for feature detection in ultrasound images is disclosed. The method estimates speckle distributions in windows on opposing sides of a pixel of an ultrasound image. The divergence is calculated for the pixel between the estimated speckle distributions in the windows. These steps are performed for each pixel in the ultrasound image, and a feature map is generated based on the divergence calculated between the estimated speckle distributions for each pixel.

Abstract: A method for lesion segmentation in 3-dimensional (3D) digital images, includes selecting a 2D region of interest (ROI) from a 3D image, the ROI containing a suspected lesion, extending borders of the ROI to 3D forming a volume of interest (VOI), where voxels on the borders of the VOI are initialized as background voxels and voxels in an interior of the VOI are initialized as foreground voxels, propagating a foreground and background voxel competition where for each voxel in the VOI, having each neighbor voxel in a neighborhood of the voxel attack the voxel, and, if the attack is successful, updating a label and strength of the voxel with that of the successful attacking voxel, and evolving a surface between the foreground and background voxels in 3D until an energy functional associated with the surface converges in value, where the surface segments the suspected lesion from the image.

Abstract: A method and system for structure enhancement and noise reduction of medical images using adaptive filtering is disclosed. The method utilizes feature estimation methods to determine multiple feature values for each pixel in an input image. Each pixel is then filtered using a filter type selected based on the feature values for that pixel.

Abstract: A method and apparatus for generating a three dimensional representation of an ear canal are disclosed whereby an ear canal of a patient is imaged using optical coherence tomography (OCT). In a first embodiment, cross-section images of an ear canal are taken by, for example, rotating an OCT imaging sensor about a predetermined axis at each of a plurality of positions. In accordance with another embodiment, a contour line is then identified in each of the cross section images and a flow algorithm is used to identify the boundary of the ear canal cross section. Once the boundaries of each cross section have been identified, all of the cross section images are combined to generate a three dimensional image of the ear canal.

Abstract: A method and system for suppressing bone structures based on a single x-ray image is disclosed. The bone structure suppressing method predicts a soft-tissue image without bone structures from an input x-ray image. A set of features is extracted for each pixel of the input x-ray image. A soft-tissue image is then generated from the input x-ray image using a trained regression function to determine an intensity value for the soft-tissue image corresponding to each pixel of the input x-ray image based on the set of features extracted for each pixel of the input x-ray image. The extracted features can be wavelet features and the regression function can be trained using Bayesian Committee Machine (BCM) to approximate Gaussian process regression (GPR).

Abstract: A method of designing hearing aid molds is disclosed whereby two shapes corresponding to graphical images of ear impressions are registered with each other to facilitate joint processing of the hearing aid design. In a first embodiment, a first graphical representation of a first ear impression is received and a feature, such as the aperture of the ear impression, is identified on that graphical model. A first vector is generated that represents the orientation and shape of that first feature. The three-dimensional translation and rotation of that first vector are determined that are necessary to align the first vector with a second vector representing the orientation and a shape of a feature, once again such as the aperture, of a second ear impression. In another embodiment, this alignment is then refined by minimizing the sum of the distances between points on the first and second graphical representations.

Abstract: A method and system for detecting the second bend plane of an ear canal surface is disclosed. A region of interest of the ear canal is defined between the first bend plane and the ear canal tip. A set of curves is defined in the region of interest and a set of high curvature points is detected from the set of curves. The second bend plane is detected using the set of high curvature points.

Abstract: A method and apparatus is disclosed whereby a point on an ear impression model to be labeled is selected and a shape context is determined for that point. This shape context is then compared to average shape contexts for different regions on a reference ear impression model, also referred to herein as an ear impression shape atlas. A cost function is used to determine the minimum cost between the shape context for the selected point and one of the average shape contexts. Once the minimized cost is determined, the region label corresponding to the average shape context having a minimized cost is assigned to that point. In this way, points on the surface of an ear impression are classified and labeled as being located in regions corresponding to the regions on the ear impression shape atlas.

Abstract: A method for rigid registration of ear impression models, including: extracting a canal region from an undetailed ear impression model, the undetailed ear impression model representing an undetailed surface of an ear canal and outer ear geometry; extracting a canal region from a detailed ear impression model, the detailed ear impression model representing a detailed surface of the ear canal; generating an orientation histogram for the canal region of the undetailed ear impression model and an orientation histogram for the canal region of the detailed ear impression model; performing a rotational alignment between the orientation histograms; computing a translational shift between the canal regions after performing the rotational alignment; and performing a registration between the undetailed and detailed ear impression models after computing the translational shift.

Abstract: An improved method of designing hearing aid molds is disclosed whereby regions of an ear impression model are identified as a function of a geodesic distance measure. According to a first embodiment, a canal point of an ear impression model is identified as that point having a maximum normalized geodesic distance as compared to all other points on the surface of the ear impression model. According to a second embodiment, a helix point of the ear impression model is identified as that point having a maximum normalized geodesic distance as compared to all points except those points in the canal region of the ear impression model. Finally, in accordance with another embodiment, a geodesic distance between a canal point and a helix point of an ear impression model is identified and a percentage threshold, illustratively 65%, is applied to that geodesic distance to identify a crus region.

Abstract: A method of designing hearing aid molds is disclosed whereby skeletons, or simplified models, of two ear impressions are used to register the graphical representations of the molds to facilitate the joint design of hearing aid shells. The center points of at least a portion of contour lines on the surface of each ear impression are identified. Then, for each ear impression, by connecting these center points to each adjacent center point, a skeleton that represents a simplified model of an ear impression is generated. Vectors describing the distance and direction from the points of each skeleton to an anatomical feature of each ear impression are identified to obtain a correspondence between the points of each skeleton. Three-dimensional translations and rotations of a feature vector of at least one of the skeletons are determined to achieve alignment of the skeleton of one ear impression with the skeleton of another impression.

Abstract: A method identifying apertures of ear impressions is disclosed. A plurality of contour lines associated with an ear impression are determined and a difference value between a value of a characteristic, such as the diameter, of each contour line and that characteristic of an adjacent contour line is determined. The aperture is identified as being that contour line having the greatest difference value. The contour lines are determined by identifying where a plane intersects the surface of the graphical representations. In another embodiment, the contour lines are assigned a weight. A contour index is then calculated for each contour line as a function of the difference value and these weights. According to this embodiment, the aperture is identified as being a contour line that is adjacent to that contour line having the greatest contour index.

Abstract: A method for modeling a 2-dimensional tubular structure in a digitized image includes providing a digitized image of a tubular structure containing a plurality of 2D balls of differing radii, initializing a plurality of connected spline segments that form an envelope surrounding the plurality of 2D balls, each the spline segment Si being parameterized by positions of the ith and i+1th balls and contact angles ?i, ?i+1 from the center of each respective ball to a point on the perimeter of each the ball contacting the spline segment Si, each the ?i affecting spline segment Si and Si?1, and updating the angles by minimizing an energy that is a functional of the angles, where the updating is repeated until the energy is minimized subject to a constraint that the envelope is tangent to each ball at each point of contact, where the envelope is represented by the contact angles.

Abstract: Disclosed is a method and system for propagating information about an image to segment a target structure. An input to denote a region of interest of the image containing the target structure is received. A first seed is set outside of the target structure and a second seed is set inside the target structure. A first partial differential equation (PDE) associated with the first seed is solved and a second PDE associated with the second seed is solved to segment the image.

Abstract: A method and system for image segmentation by evolving radial basis functions (RBFs) is disclosed. A set of RBFs define a contour on an image. The contour is the zero level set of an implicit function defined by the RBFs. The RBFs are weighted and parameterized by multiple parameters, such as center point, standard deviation, and orientation. The contour is iteratively deformed by updating the weight and each of the parameters of each of the RBFs based on each pixel of the contour, until the contour converges. The parameters and the weight of each RBF is updated using coupled set of differential equations derived based from a region-based or boundary-based image segmentation energy formulation. The final contour at a convergence defines boundaries of a target object or region in the image.