Methods, devices, apparatus, and systems for three-dimensional (3D) contoured scanning photoacoustic imaging and/or deep learning virtual staining.

A mobile phone-based miniature microscopic image acquisition device, and image stitching and recognition methods are provided. The acquisition device comprises a support, wherein a mobile phone fixing table is provided on the support. A microscope head is provided below a camera of a mobile phone. A slide holder is provided below the microscope head, and an lighting source is provided below the slide holder. A scanning movement is performed between the slide holder and the microscope head along X and Y axes, so that images of a slide are acquired into the mobile phone. The slide sample images acquired into the mobile phone can be stitched and recognized, and can be uploaded to the cloud to be processed by cloud AI, thereby significantly improving the accuracy and efficiency of cell recognition, greatly reducing the medical cost, and ensuring more remote medical institutions can apply such technology for diagnosis.

A urine analysis system according to an embodiment includes: a testing apparatus that measures particles included in a urine sample according to a flow cytometry method; an image capturing apparatus that captures images of particles in the urine sample to acquire particle images; and a management apparatus that receives a measurement result obtained by the testing apparatus and the particle images acquired by the image capturing apparatus. The management apparatus generates an order to capture an image of the urine sample based on the measurement result obtained by the testing apparatus. The image capturing apparatus executes the image capturing processing of the particles in the urine sample for which the image capturing order has been generated by the management apparatus, and transmits the acquired particle images to the management apparatus.

A classification training method for training classifiers adapted to identify specific mutations associated with different cancer including identifying driver mutations. First cells from mutation cell lines derived from conditions having the number of driver mutations are acquired and 3D image feature data from the number of first cells is identified. 3D cell imaging data from the number of first cells and from other malignant cells is generated, where cell imaging data includes a number of first individual cell images. A second set of 3D cell imaging data is generated from a set of normal cells where the number of driver mutations are expected to occur, where the second set of cell imaging data includes second individual cell images. Supervised learning is conducted based on cell line status as ground truth to generate a classifier.

Disclosed is a computer-implemented method which encompasses registering a tracked imaging device such as a microscope having a known viewing direction and an atlas to a patient space so that a transformation can be established between the atlas space and the reference system for defining positions in images of an anatomical structure of the patient. Labels are associated with certain constituents of the images and are input into a learning algorithm such as a machine learning algorithm, for example a convolutional neural network, together with the medical images and an anatomical vector and for example also the atlas to train the learning algorithm for automatic segmentation of patient images generated with the tracked imaging device. The trained learning algorithm then allows for efficient segmentation and/or labelling of patient images without having to register the patient images to the atlas each time, thereby saving on computational effort.

Automation of microscopic pathological diagnosis relies on digital image quality, which, in turn, affects the rates of false positive and negative cellular objects designated as abnormalities. Cytogenetic biodosimetry is a genotoxic assay that detects dicentric chromosomes (DCs) arising from exposure to ionizing radiation. The frequency of DCs is related to radiation dose received, so the inferred radiation dose depends on the accuracy of DC detection. To improve this accuracy, image segmentation methods are used to rank high quality cytogenetic images and eliminate suboptimal metaphase cell data in a sample based on novel quality measures. When sufficient numbers of high quality images are found, the microscope system is directed to terminate metaphase image collection for a sample. The International Atomic Energy Agency recommends at least 500 images be used to estimate radiation dose, however often many more images are collected in order to select the metaphase cells with good morphology for analysis. Improvements in DC recognition increase the accuracy of dose estimates, by reducing false positive (FP) DC detection. A set of chromosome morphology segmentation methods selectively filtered out false DCs, arising primarily from extended prometaphase chromosomes, sister chromatid separation and chromosome fragmentation. This reduced FPs by 55% and was highly specific to the abnormal structures (≥97.7%). Additional procedures were then developed to fully automate image review, resulting in 6 image-level filters that, when combined, selectively remove images with consistently unparsable or incorrectly segmented chromosome morphologies. Overall, these filters can eliminate half of the FPs detected by manual image review. Optimal image selection and FP DCs are minimized by combining multiple feature based segmentation filters and a novel image sorting procedure based on the known distribution of chromosome lengths. Consequently, the average dose estimation error was reduced from 0.4 Gy to <0.2 Gy with minimal manual review required. Automated image selection with these filters reduces the number of images that are required to capture metaphase cells, thus decreasing the number of images and time required for each sample. A microscope system integrates image selection procedures controls with an automated digitally controlled microscope then determines at what point a sufficient number of metaphase cell images have been acquired to accurately determine radiation dose, which then terminates data collection by the microscope. These image filtering approaches constitute a reliable and scalable solution that results in more accurate and rapid radiation dose es

A facility for identifying the boundaries of 3-dimensional structures in 3-dimensional images is described. For each of multiple 3-dimensional images, the facility receives results of a first attempt to identify boundaries of structures in the 3-dimensional image, and causes the results of the first attempt to be presented to a person. For each of a number of 3-dimensional images, the facility receives input generated by the person providing feedback on the results of the first attempt. The facility then uses the following to train a deep-learning network to identify boundaries of 3-dimensional structures in 3-dimensional images: at least a portion of the plurality of 3-dimensional images, at least a portion of the received results, and at least a portion of provided feedback.

Systems and methods for spatial analysis of analytes are provided. A data structure is obtained comprising an image, as an array of pixel values, of a sample on a substrate having a identifier, fiducial markers and a set of capture spots. The pixel values are used to identify derived fiducial spots. The substrate identifier identifies a template having reference positions for reference fiducial spots and a corresponding coordinate system. The derived fiducial spots are aligned with the reference fiducial spots using an alignment algorithm to obtain a transformation between the derived and reference fiducial spots. The transformation and the template corresponding coordinate system are used to register the image to the set of capture spots. The registered image is then analyzed in conjunction with spatial analyte data associated with each capture spot, thereby performing spatial analysis of analytes.

A method comprising receiving images depicting stained target tissue, segmenting the images into cell type and region type segmentations, extracting cell phenotype features from an analysis of the stains for cell type segmentations, clustering the cell type segmentations, computing feature vectors each including the respective cell phenotype features, and an indication of a location of the cell type segmentation relative to region type segmentation(s), creating a cell-graph based on the feature vectors of cell type segmentations and/or clusters, wherein each node denotes respective cell type segmentation and/or respective cluster and includes the feature vector, and edges represent a physical distance between cell type segmentations and/or clusters corresponding to the respective nodes, inputting the cell-graph into a graph neural network, and obtaining an indication of a target therapy likely to be effective for treatment of medical condition in the subject as an outcome of the graph neural network.

A way to design a codebook for estimating the type of a molecule at a particular location in a fluorescence microscopy image makes use of one or both of (1) knowledge of the non-uniform prior distribution of molecule types (i.e., some types are known a priori to occur more frequently than others) and/or knowledge of co-occurrence of molecule types at close locations (e.g., in a same cell); and (2) knowledge of a model of the (e.g., random) process that yields the intensities that are expected at a location when a molecule with a particular subset of markers (i.e., a molecule of a type that has been assigned a codeword that defines that subset) is present at that location. The codebook design may provide experimental efficiency by reducing the number of images that need to be acquired and/or improve classification or detection accuracy by making the codewords for different molecule types more distinctive.