Abstract: Labeling protocols for training datasets and systems and methods for classifying objects of interest and identifying backgrounds. Labeling protocols described herein enable grid units within an image to be excluded from contributing to a loss function, and further enable granular controls for specifying which grid units should contribute to object predictions. A method partitioning an input image into a plurality of grid units and individually processing each of the plurality of grid units with a neural network configured to calculate a confidence score indicating a likelihood that a grid unit comprises an object of interest. The method includes generating a bounding box around the object of interest, identifying one or more grid units of the plurality of grid units that comprise a portion of the bounding box, and identifying which of the one or more grid units comprises a center point of the bounding box.

Abstract: Machine learning image analysis for quantitative and qualitative analysis of agglutination samples. A method includes receiving an image of an agglutination assay comprising a negative control sample, a positive control sample, and a test sample. The method includes providing the image to a machine learning algorithm trained to classify agglutination of the test sample on a quantitative scale. The machine learning algorithm calibrates the quantitative scale based at least in part on the negative control sample and the positive control sample.

Abstract: Protecting information by partitioning data across multiple storage resources. A method includes receiving an image of a test card and an identity of a patient associated with the test card. The method includes associating the image of the test card with the identity of the patient as a data pair on a first storage resource. The method includes providing the image of the test card to a second storage resource by way of a network, wherein the first storage resource is independent of the second storage resource. The method includes receiving a result for the test card and associating the result for the test card with the identity of the patient based on the data pair stored on the first storage resource.

Abstract: Test cards for agglutination assays, wherein the test cards are configured for computer-implemented image analysis. A test card includes a negative control test region for receiving a negative control fluid sample, a positive control test region for receiving a positive control fluid sample, and a test sample region for receiving a test fluid sample. The test card includes one or more unique scannable codes comprising data for instructing a processor to capture an image of the test card that is suitable for computer-implemented image analysis.

Abstract: Machine learning analysis for classifying agglutination of fluid samples. A method includes scanning a unique scannable code printed on a test card, wherein the test card comprises a negative control fluid sample, a positive control fluid sample, and a test fluid sample. The method includes capturing an image of the test card and providing the image of the test card to a machine learning algorithm configured to assess agglutination of the test fluid sample based on the image. The method includes receiving from the machine learning algorithm one or more of a qualitative analysis or a quantitative analysis of the agglutination of the test fluid sample.

Abstract: Systems, methods, and devices for identifying a reference focal plane in connection with light microscopy. A method includes identifying a fiducial marker printed on a surface of a coverslip or a slide used in connection with optical microscopy. The method includes focusing the optical microscope on the fiducial marker to calculate a focal distance of the fiducial marker. The method includes calculating a reference focal plane defining the surface of the coverslip or the slide based on the focal distance of the fiducial marker.

Abstract: Systems, methods, and devices for identifying a reference focal plane in connection with light microscopy. A method includes identifying a fiducial marker printed on a surface of a coverslip or a slide used in connection with optical microscopy. The method includes focusing the optical microscope on the fiducial marker to calculate a focal distance of the fiducial marker. The method includes calculating a reference focal plane defining the surface of the coverslip or the slide based on the focal distance of the fiducial marker.

Abstract: The disclosure relates to machine learning classification of cells/particles in microscopy images. A method includes inputting an image having invisible features into an initial neural network classifier (INNC) of a convolutional neural network. The INNC is trained using images with ground truth derived from out-of-channel mechanisms. The method includes generating an intermediate classification from the original image. The intermediate classification and the original image are input into a final neural network classifier (FNNC) that comprises one or more bypass layers to feed forward an initial, final classification from a final activation layer to a final convolutional layer thereby bypassing a final pooling layer. The final convolutional layer has an increased kernel size and more filters than the initial convolutional layer. The final classification is generated based on the invisible features in the original image and outputted.

Abstract: Systems, methods, and devices for classifying or detecting mold samples or training computer models (such as neural networks), are disclosed. A method includes obtaining a microscopy image of a mold sample. The method includes determining a classification of the mold sample based on non-image data corresponding to the mold sample. The method further includes training a computer model based on the microscopy image and a label indicating the classification.

Abstract: Systems, methods, and devices for training models or algorithms for classifying or detecting particles or materials in microscopy images are disclosed. A method includes receiving a plurality of microscopy images of a specimen and a classification for the specimen. The plurality of microscopy images includes a first image captured at a first magnification and a second image captured at the first magnification with a different focus than the first image. The method includes training a machine learning model or algorithm using the plurality of images, wherein the first image and the second image are provided with one or more labels indicating the classification.

Abstract: Systems, methods, and devices for training models or algorithms for classifying or detecting particles or materials in microscopy images are disclosed. A method includes receiving a plurality of microscopy images of a specimen and a classification for the specimen. The plurality of microscopy images includes a first image captured at a first magnification and a second image captured at the first magnification with a different focus than the first image. The method includes training a machine learning model or algorithm using the plurality of images, wherein the first image and the second image are provided with one or more labels indicating the classification.

Abstract: Systems, methods, and devices for classifying or detecting mold samples or training computer models (such as neural networks), are disclosed. A method includes obtaining a microscopy image of a mold sample. The method includes determining a classification of the mold sample based on non-image data corresponding to the mold sample. The method further includes training a computer model based on the microscopy image and a label indicating the classification.

Abstract: The disclosure relates to machine learning classification of cells/particles in microscopy images. A method includes inputting an image having invisible features into an initial neural network classifier (INNC) of a convolutional neural network. The INNC is trained using images with ground truth derived from out-of-channel mechanisms. The method includes generating an intermediate classification from the original image. The intermediate classification and the original image are input into a final neural network classifier (FNNC) that comprises one or more bypass layers to feed forward an initial, final classification from a final activation layer to a final convolutional layer thereby bypassing a final pooling layer. The final convolutional layer has an increased kernel size and more filters than the initial convolutional layer. The final classification is generated based on the invisible features in the original image and outputted.