The technology relates to software, systems, and methods for automated medical image segmentation via deep learning. Standard holistically-nested networks (HNNs) used in computer vision and medical imaging for segmentation and edge detection have a problem with coarsening resolution. The described technology addresses this.
Accurate automated organ and disease feature segmentation is a challenge for medical imaging analysis. The pancreas, for example, is a small, soft, organ with low uniformity of shape and volume between patients. Because of the lack of uniform image patterns, there are few features that can be used to aid in automated identification of anatomy and boundaries. Segmentation of high variability features is uniquely difficult for a computer to perform.
Multiparametric MRI improves image detail and prostate cancer detection rates compared to standard MRI. Computer aided diagnostics (CAD) used in combination with multiparametric MRI images may further improve prostate cancer detection and visualization. The technology, developed by researchers at the National Institutes of Health Clinical Center (NIHCC), is an automated CAD system for use in processing and visualizing prostate lesions on multiparametric MRI images.
Accurate automated organ and disease feature segmentation is a challenge for medical imaging analysis. The pancreas, for example, is a small, soft, organ with low uniformity of shape and volume between patients. Because of the lack of uniform image patterns, there are few features that can be used to aid in automated identification of anatomy and boundaries. Segmentation of high variability features is uniquely difficult for a computer to perform.
Developing computer systems that can recognize and locate image features associated with disease is a challenge for developing fully-automated and high precision computer assisted diagnostics. Joint learning of language tasks in association with vision tasks (association of image features with text annotation) adds an additional level of challenge. Furthermore, scaling-up approaches from small to large datasets presents additional issues, particularly related to medical images.
The successful treatment of cancer is correlated with the early detection of the cancerous cells. Conventional cancer diagnosis is largely based on qualitative morphological criteria, but more accurate quantitative tests could greatly increase early detection of malignant cells. It has been observed that the spatial arrangement of DNA in the nucleus is altered in cancer cells in comparison to normal cells. Therefore, it is possible to distinguish malignant cells by mapping the position of labeled marker genes in the nucleus.
Available for commercial development is software that provides automatic visualization of features inside biological image volumes in 3D. The software provides a simple and interactive visualization for the exploration of biological datasets through dataset-specific transfer functions and direct volume rendering. The method employs a K-Means++ clustering algorithm to classify a two-dimensional histogram created from the input volume. The classification process utilizes spatial and data properties from the volume.
The National Cancer Institute (NCI) seeks parties to license software for modeling the targeted delivery of anti-cancer agents in solid tumors.
The software models the permeability and concentration of intravenously administered antibody anti-cancer agent conjugates in solid tumors. The models can be used to determine optimal dosing regimen of a therapeutic in a particular cancer type. Thirty factors that affect delivery rates and efficiencies are analyzed as variables in generating the models.
Accurate measurement of low metabolite concentrations produced by medically important enzymes is commonly obscured by noise during magnetic resonance imaging (MRI). Measuring the turnover rate of low-level metabolites can directly quantify the activity of enzymes of interest, including possible drug targets in cancer and other diseases. Noise can cause the in vivo signal to fall below the limit of detection. A variety of denoising methods have been proposed to enhance spectroscopic peaks, but still fall short for the detection of low-intensity signals.
This invention pertains to a system for continuous observation of rodents in home-cage environments with the specific aim to facilitate the quantification of activity levels and behavioral patterns for mice housed in a commercial ventilated cage rack. The home-cage in-rack provides daytime and nighttime monitoring with the stability and consistency of a home cage environment.