Abstract: Systems and methods for using a prediction model jointly with a normalization model to provide prediction results are provided. One example method includes receiving an input image of a tissue sample of a patient and generating a normalized image by applying a normalization model on the input image. The normalization model is configured to generate normalized data using input data for a prediction model, and the prediction model is configured to generate prediction results using normalized data generated by the normalization model. The normalization model and the prediction model are jointly trained. The method further includes generating a prediction of disease severity for the patient by applying the prediction model on the normalized image.

Abstract: Systems and methods for predicting images with enhanced spatial resolution using a neural network are provided herein. According to an aspect of the invention, a method includes accessing an input image of a biological sample, wherein the input image includes a first spatial resolution and a plurality of spectral images, and wherein each spectral image of the plurality of spectral images includes data from a different wavelength band at a different spectral channel; applying a trained artificial neural network to the input image; generating an output image at a second spatial resolution, wherein the second spatial resolution is higher than the first spatial resolution, and wherein the output image includes a fewer number of spectral channels than the plurality of spectral images included in the input image; and outputting the output image.

Abstract: Systems and methods of improving alignment in dense prediction neural networks are disclosed. A method includes identifying, at a computing system, an input data set and a label data set with one or more first parts of the input data set corresponding to a label. The computing system processes the input data set using a neural network to generate a predicted label data set that identifies one or more second parts of the input data set predicted to correspond to the label. The computing system determines an alignment result using the predicted label data set and the label data set and a transformation of the one or more first parts, including a shift, rotation, scaling, and/or deformation, based on the alignment result. The computing system computes a loss score using the transformation, label data and the predicted label data set and updates the neural network based on the loss score.

Abstract: Various techniques are provided for performing automated full-cell segmentation and labeling in immunofluorescent microscopy. These techniques perform membrane segmentation and nuclear seed detection separate and independently from each other, then combine their results to identify cell boundaries. Some embodiments use texture- and kernel-based image processing to perform the method. In some embodiments, the method for obtaining membrane features disclosed herein can be used in conjunction with or separate from the nuclear features. The results can be used for a variety of purposes, including whole-area cell segmentation in fluorescence-based tissue imaging.

Abstract: Systems and methods for predicting images with enhanced spatial resolution using a neural network are provided herein. According to an aspect of the invention, a method includes accessing an input image of a biological sample, wherein the input image includes a first spatial resolution and a plurality of spectral images, and wherein each spectral image of the plurality of spectral images includes data from a different wavelength band at a different spectral channel; applying a trained artificial neural network to the input image; generating an output image at a second spatial resolution, wherein the second spatial resolution is higher than the first spatial resolution, and wherein the output image includes a fewer number of spectral channels than the plurality of spectral images included in the input image; and outputting the output image.

Abstract: Presented herein are systems and methods that allow for vertebral centrums of individual vertebrae to be identified and segmented within a 3D image of a subject (e.g., a CT or microCT image). In certain embodiments, the approaches described herein identify, within a graphical representation of an individual vertebra in a 3D image of a subject, multiple discrete and differentiable regions, one of which corresponds to a vertebral centrum of the individual vertebra. The region corresponding to the vertebral centrum may be automatically or manually (e.g., via a user interaction) classified as such. Identifying vertebral centrums in this manner facilitates streamlined quantitative analysis of 3D images for osteological research, notably, providing a basis for rapid and consistent evaluation of vertebral centrum morphometric attributes.

Abstract: Systems and methods of improving alignment in dense prediction neural networks are disclosed. A method includes identifying, at a computing system, an input data set and a label data set with one or more first parts of the input data set corresponding to a label. The computing system processes the input data set using a neural network to generate a predicted label data set that identifies one or more second parts of the input data set predicted to correspond to the label. The computing system determines an alignment result using the predicted label data set and the label data set and a transformation of the one or more first parts, including a shift, rotation, scaling, and/or deformation, based on the alignment result. The computing system computes a loss score using the transformation, label data and the predicted label data set and updates the neural network based on the loss score.

Abstract: Presented herein are efficient and reliable systems and methods for calculating and extracting three-dimensional central axes of bones of animal subjects—for example, animal subjects scanned by in vivo or ex vivo microCT platforms—to capture both the general and localized tangential directions of the bone, along with its shape, form, curvature, and orientation. With bone detection and segmentation algorithms, the skeletal bones of animal subjects scanned by CT or microCT scanners can be detected, segmented, and visualized. Three dimensional central axes determined using these methods provide important information about the skeletal bones.

Abstract: Systems and methods of improving alignment in dense prediction neural networks are disclosed. A method includes identifying, at a computing system, an input data set and a label data set with one or more first parts of the input data set corresponding to a label. The computing system processes the input data set using a neural network to generate a predicted label data set that identifies one or more second parts of the input data set predicted to correspond to the label. The computing system determines an alignment result using the predicted label data set and the label data set and a transformation of the one or more first parts, including a shift, rotation, scaling, and/or deformation, based on the alignment result. The computing system computes a loss score using the transformation, label data and the predicted label data set and updates the neural network based on the loss score.

Abstract: Presented herein are systems and methods that allow for vertebral centrums of individual vertebrae to be identified and segmented within a 3D image of a subject (e.g., a CT or microCT image). In certain embodiments, the approaches described herein identify, within a graphical representation of an individual vertebra in a 3D image of a subject, multiple discrete and differentiable regions, one of which corresponds to a vertebral centrum of the individual vertebra. The region corresponding to the vertebral centrum may be automatically or manually (e.g., via a user interaction) classified as such. Identifying vertebral centrums in this manner facilitates streamlined quantitative analysis of 3D images for osteological research, notably, providing a basis for rapid and consistent evaluation of vertebral centrum morphometric attributes.

Abstract: Presented herein are systems and methods for tomographic imaging of a region of interest in a subject using short-wave infrared light to provide for accurate reconstruction of absorption maps within the region of interest. The reconstructed absorption maps are representations of the spatial variation in tissue absorption within the region of interest. The reconstructed absorption maps can themselves be used to analyze anatomical properties and biological processes within the region of interest, and/or be used as input information about anatomical properties in order to facilitate data processing used to obtain images of the region of interest via other imaging modalities. For example, the reconstructed absorption maps may be incorporated into forward models that are used in tomographic reconstruction processing in fluorescence and other contrast-based tomographic imaging modalities.

Abstract: Presented herein are systems and methods that provide for fast image acquisition with a CCD camera for tomographic imaging by synchronizing illumination with the image acquisition sequence of the CCD camera. The systems and methods described herein allow images to be acquired with a CCD camera using short image acquisition times that would otherwise result in the introduction of severe artifacts into the acquired images. This unique capability is achieved by selectively illuminating the one or more object(s) to be imaged during specific phases of the CCD camera that are used to acquire an image. Reducing the time required to acquire artifact-free images in this manner allows for rapid imaging with a CCD camera. This capability is of particular relevance to tomographic imaging approaches, in which multiple images of one or more objects are acquired and used to produce a single tomographic image.

Abstract: Presented herein are systems and methods for tomographic imaging that provide for rapid illumination of multiple excitation locations across a large field of view by one or more beams of excitation light from one or more excitation sources. The approaches described herein utilize a galvanometer optical scanner to scan a beam of excitation light through a plurality of locations across a scan region corresponding to the field of view to be imaged. In certain embodiments, the systems and methods described herein utilize beams of excitation light with specifically tailored shapes to maintain small spot sizes across the large scan region. The ability to scan over a large region while still maintaining small spot sizes provided by the approaches described herein allows for accurate, high-resolution tomographic imaging of large or multiple subjects, thereby expanding the capabilities of tomographic imaging systems.

Abstract: Systems and methods of improving alignment in dense prediction neural networks are disclosed. A method includes identifying, at a computing system, an input data set and a label data set with one or more first parts of the input data set corresponding to a label. The computing system processes the input data set using a neural network to generate a predicted label data set that identifies one or more second parts of the input data set predicted to correspond to the label. The computing system determines an alignment result using the predicted label data set and the label data set and a transformation of the one or more first parts, including a shift, rotation, scaling, and/or deformation, based on the alignment result. The computing system computes a loss score using the transformation, label data and the predicted label data set and updates the neural network based on the loss score.

Abstract: Presented herein are systems and methods that facilitate automated segmentation of 3D images of subjects to distinguish between regions of heterotopic ossification (HO) normal skeleton, and soft tissue. In certain embodiments, the methods identify discrete, differentiable regions of a 3D image of subject (e.g., a CT or microCT image) that may then be either manually or automatically classified as either HO or normal skeleton.

Abstract: Various techniques are provided for performing automated full-cell segmentation and labeling in immunofluorescent microscopy. These techniques perform membrane segmentation and nuclear seed detection separate and independently from each other, then combine their results to identify cell boundaries. Some embodiments use texture- and kernel-based image processing to perform the method. In some embodiments, the method for obtaining membrane features disclosed herein can be used in conjunction with or separate from the nuclear features. The results can be used for a variety of purposes, including whole-area cell segmentation in fluorescence-based tissue imaging.

Abstract: Presented herein are efficient and reliable systems and methods for calculating and extracting three-dimensional central axes of bones of animal subjects—for example, animal subjects scanned by in vivo or ex vivo microCT platforms—to capture both the general and localized tangential directions of the bone, along with its shape, form, curvature, and orientation. With bone detection and segmentation algorithms, the skeletal bones of animal subjects scanned by CT or microCT scanners can be detected, segmented, and visualized. Three dimensional central axes determined using these methods provide important information about the skeletal bones.

Abstract: Presented herein are efficient and reliable systems and methods for calculating and extracting three-dimensional central axes of bones of animal subjects—for example, animal subjects scanned by in vivo or ex vivo microCT platforms—to capture both the general and localized tangential directions of the bone, along with its shape, form, curvature, and orientation. With bone detection and segmentation algorithms, the skeletal bones of animal subjects scanned by CT or microCT scanners can be detected, segmented, and visualized. Three dimensional central axes determined using these methods provide important information about the skeletal bones.

Abstract: Presented herein are systems and methods that provide for fast image acquisition with a CCD camera for tomographic imaging by synchronizing illumination with the image acquisition sequence of the CCD camera. The systems and methods described herein allow images to be acquired with a CCD camera using short image acquisition times that would otherwise result in the introduction of severe artifacts into the acquired images. This unique capability is achieved by selectively illuminating the one or more object(s) to be imaged during specific phases of the CCD camera that are used to acquire an image. Reducing the time required to acquire artifact-free images in this manner allows for rapid imaging with a CCD camera. This capability is of particular relevance to tomographic imaging approaches, in which multiple images of one or more objects are acquired and used to produce a single tomographic image.

Abstract: Presented herein are efficient and reliable systems and methods for calculating and extracting three-dimensional central axes of bones of animal subjects—for example, animal subjects scanned by in vivo or ex vivo microCT platforms—to capture both the general and localized tangential directions of the bone, along with its shape, form, curvature, and orientation. With bone detection and segmentation algorithms, the skeletal bones of animal subjects scanned by CT or microCT scanners can be detected, segmented, and visualized. Three dimensional central axes determined using these methods provide important information about the skeletal bones.