Abstract: A system and method of multi-stage training of a transformer-based machine-learning model. The system performs at least two stages of the following three stages of training: During a first stage, the system pre-trains a transformer encoder via a first machine-learning network using an unlabeled 3D image dataset. During a second stage, the system fine-tunes the pre-trained transformer encoder via a second machine-learning network via a labeled 2D image dataset. During a third stage, the system further fine-tunes the previously pre-trained transformer encoder or fine-tuned transformer encoder via a third machine-learning network using a labeled 3D image dataset.

Abstract: A system and method are disclosed for evaluating neural network prediction by leveraging generative reconstructions of a test image to assess confidence in the prediction. The system obtains a test image of a tissue histology slide. The system generates seeds from applying a dropout filter to the test image. The system generates, for each seed, a synthetic image by applying an image generator model to the seed. The system applies a prediction model (e.g., a neural network) to the test image to generate a prediction for the test image of whether the tissue is normal or anormal. The system applies the prediction model to each synthetic image to generate a prediction for the synthetic image of whether the tissue is normal or anormal. The system determines a final prediction for the test image and a confidence associated with the final prediction based on the predictions for the test image and the synthetic images.

Abstract: A method or system for training a convolutional neural network (CNN) for medical imaging analysis. The system pre-trains the CNN's encoder using a dataset of unlabeled 3D medical images. Each 3D image includes an annotated slice delineating a boundary of a lesion and multiple non-annotated 2D slices above and below the annotated slice. The system then fine-tunes the pre-trained encoder using an annotated 2D image dataset. The annotated 2D image dataset includes multiple 2D slices of lesions, each including an annotation that delineates a boundary of a corresponding lesion.

Abstract: Methods and systems are described for image segmentation. A machine learning model is applied to a set of images to generate results. The results may be obtained as a probability map for each image in the set of images. The model may be trained by accessing a set of labeled images, each image associated with a label indicating a location of a feature within a respective image. An initial set of parameters is accessed. An encoder is initialized with the initial set of parameters. The encoder is applied to the set of labeled images to generate a prediction of a feature location within each image. The initial set of parameters are updated based on the predictions and the label associated with the labeled images. The updated set of parameters and an additional set of parameters generated using a set of unlabeled images are aggregated.

Abstract: A method or system for training a convolutional neural network (CNN) for medical imaging analysis. The system pre-trains the CNN's encoder using a dataset of unlabeled 3D medical images. Each 3D image includes an annotated slice delineating a boundary of a lesion and multiple non-annotated 2D slices above and below the annotated slice. The system then fine-tunes the pre-trained encoder using an annotated 2D image dataset. The annotated 2D image dataset includes multiple 2D slices of lesions, each including an annotation that delineates a boundary of a corresponding lesion.

Abstract: A system and method of multi-stage training of a transformer-based machine-learning model. The system performs at least two stages of the following three stages of training: During a first stage, the system pre-trains a transformer encoder via a first machine-learning network using an unlabeled 3D image dataset. During a second stage, the system fine-tunes the pre-trained transformer encoder via a second machine-learning network via a labeled 2D image dataset. During a third stage, the system further fine-tunes the previously pre-trained transformer encoder or fine-tuned transformer encoder via a third machine-learning network using a labeled 3D image dataset.

Abstract: Embodiments disclosed herein relate to a model for predicting the release profile of a controlled release device. The implant modeling system and models disclosed herein allow the accurate prediction of a release profile for a controlled release device based on features extracted from micro-resolution imagery. The models combine microstructural features that can be extracted at the XRCT resolution, including pore volume and connectivity, using erosion-dilation image analysis. This strategy allows prediction of release curves of the controlled release device using XRCT despite its resolution limitations.

Abstract: A system and method are disclosed for segmenting a set of two-dimensional CT slices corresponding to a lesion. In an embodiment, for each of at least a subset of the set of CT slices, the system inputs the CT slice into a plurality of branches of a trained segmentation block. Each branch of the segmentation block includes a convolutional neural network (CNN) with filters at a different scale, and produces a plurality of levels of output. The system generates, for each CT slice in the subset, feature maps for each level of output. The system generates a segmentation of each CT slice in the subset based on the feature maps of each level of output. The system aggregates the segmentations of each slice in the subset to generate a three-dimensional segmentation of the lesion. The system provides data representing the three-dimensional segmentation for display.

Abstract: Methods and systems are described for image segmentation. A machine learning model is applied to a set of images to generate results. The results may be obtained as a probability map for each image in the set of images. The model may be trained by accessing a set of labeled images, each image associated with a label indicating a location of a feature within a respective image. An initial set of parameters is accessed. An encoder is initialized with the initial set of parameters. The encoder is applied to the set of labeled images to generate a prediction of a feature location within each image. The initial set of parameters are updated based on the predictions and the label associated with the labeled images. The updated set of parameters and an additional set of parameters generated using a set of unlabeled images are aggregated.

Abstract: A system and method are disclosed for segmenting a set of two-dimensional CT slices corresponding to a lesion. In an embodiment, for each of at least a subset of the set of CT slices, the system inputs the CT slice into a plurality of branches of a trained segmentation block. Each branch of the segmentation block includes a convolutional neural network (CNN) with filters at a different scale, and produces a plurality of levels of output. The system generates, for each CT slice in the subset, feature maps for each level of output. The system generates a segmentation of each CT slice in the subset based on the feature maps of each level of output. The system aggregates the segmentations of each slice in the subset to generate a three-dimensional segmentation of the lesion. The system provides data representing the three-dimensional segmentation for display.

Abstract: A system and method are disclosed for segmenting a set of two-dimensional CT slices corresponding to a lesion. In an embodiment, for each of at least a subset of the set of CT slices, the system inputs the CT slice into a plurality of branches of a trained segmentation block. Each branch of the segmentation block includes a convolutional neural network (CNN) with filters at a different scale, and produces one or more levels of output. The system generates, for each CT slice in the subset, feature maps for each level of output. The system generates a segmentation of each CT slice in the subset based on the feature maps of each level of output. The system aggregates the segmentations of each slice in the subset to generate a three-dimensional segmentation of the lesion. The system transmits data representing the three-dimensional segmentation to a user interface for display.

Abstract: Approaches for the automatic segmentation of magnetic resonance (MR) images. Machine learning models segment images to identify image features in consecutive frames at different levels of resolution. A neural network block is applied to groups of MR images to produce primary feature maps at two or more levels of resolution. The images in a given group of MR images may correspond to a cycle and have a temporal order. A second RNN block is applied to the primary feature maps to produce two or more output tensors at corresponding levels of resolution. A segmentation block is applied to the two or more output tensors to produce a probability map for the MR images. The first neural network block may be a convolutional neural network (CNN) block. The second neural network block may be a convolutional long short-term (LSTM) block.

Abstract: A system and method are disclosed for segmenting a set of two-dimensional CT slices corresponding to a lesion. In an embodiment, for each of at least a subset of the set of CT slices, the system inputs the CT slice into a plurality of branches of a trained segmentation block. Each branch of the segmentation block includes a convolutional neural network (CNN) with filters at a different scale, and produces one or more levels of output. The system generates, for each CT slice in the subset, feature maps for each level of output. The system generates a segmentation of each CT slice in the subset based on the feature maps of each level of output. The system aggregates the segmentations of each slice in the subset to generate a three-dimensional segmentation of the lesion. The system transmits data representing the three-dimensional segmentation to a user interface for display.

Abstract: Approaches for the automatic segmentation of magnetic resonance (MR) images. Machine learning models segment images to identify image features in consecutive frames at different levels of resolution. A neural network block is applied to groups of MR images to produce primary feature maps at two or more levels of resolution. The images in a given group of MR images may correspond to a cycle and have a temporal order. A second RNN block is applied to the primary feature maps to produce two or more output tensors at corresponding levels of resolution. A segmentation block is applied to the two or more output tensors to produce a probability map for the MR images. The first neural network block may be a convolutional neural network (CNN) block. The second neural network block may be a convolutional long short-term (LSTM) block.

Abstract: Embodiments disclosed herein relate to a model for predicting the release profile of a controlled release device. The implant modeling system and models disclosed herein allow the accurate prediction of a release profile for a controlled release device based on features extracted from micro-resolution imagery. The models combine microstructural features that can be extracted at the XRCT resolution, including pore volume and connectivity, using erosion-dilation image analysis. This strategy allows prediction of release curves of the controlled release device using XRCT despite its resolution limitations.