Methods of assessing lung disease in chest X-rays
Inventors
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-12327356-B2
Publication Date
2025-06-10
Expiration Date
Abstract
The present system provides methods and systems of detecting lung abnormalities in chest x-ray images using at least two neural networks.
Core Innovation
The invention relates to a diagnostic method that provides an image file of a chest x-ray from a patient to a machine learning system trained on training data only available at a plurality of sources separated by time and/or geography. The training connects the machine learning system to the plurality of sources at different times and/or locations, so that the machine learning system learns associations between features in chest x-rays and known pathology results.
The machine learning system resizes the image file of the chest x-ray into a first image that depicts the entire x-ray at a reduced resolution and places a subsection of the file into a second image at an original resolution. The method analyzes the first and second images in parallel by respective first and second neural networks to output scores indicating a probability of a nodule.
The invention further supports privacy-preserving training by using training data only available at multiple sources separated by time and/or geography. Documented aspects include CT-based ground truth and federated/distributed learning, and workflows for preprocessing and lung segmentation before nodule detection, together with detection and localization outputs such as bounding boxes and heatmap-style region indications.
Claims Coverage
The partial content provides one independent claim (clm-00001). It covers a diagnostic method for lung nodule detection that combines privacy-preserving training from multiple time/geography-separated sources with a two-scale parallel neural network analysis pipeline and performance constrained by an AUC range.
Federated training from time and/or geography separated sources
providing an image file of a chest x-ray from a patient to a machine learning system that has been trained on training data only available at a plurality of sources separated by time and/or geography, and the training comprises connecting the machine learning system to the plurality of sources at different times and/or locations
Two-scale parallel neural analysis of entire reduced-resolution image and original-resolution subsection
resizing the image file of the chest x-ray into a first image that depicts the entire x-ray at a reduced resolution and placing a subsection of the file into a second image at an original resolution, and analyzing the first and second images in parallel by respective first and second neural networks to output scores indicating a probability of a nodule
Performance constrained learned feature associations for nodule detection (AUC range)
trained to learn associations between features in chest x-rays and known pathology results with an area under the curve (AUC) of true positives over false positives for learned feature associations between 0.7 and 0.83
Operating the machine learning system to detect lung nodules
operating the machine learning system to detect lung nodules
Across the independent claim, the core coverage centers on training using data only available across multiple time and/or geography-separated sources, parallel analysis of a reduced-resolution full-view image and an original-resolution subsection using first and second neural networks to output nodule probability scores, and learned feature associations tied to known pathology results with an AUC range between 0.7 and 0.83, culminating in detection of lung nodules.
Stated Advantages
Not explicitly described in patent.
Documented Applications
Not explicitly described in patent.
Interested in licensing this patent?