METHODS OF ASSESSING LUNG DISEASE IN CHEST X-RAYS
The present system provides methods and systems of detecting lung abnormalities in chest x-ray images using at least two neural networks.
1 .- 20 . (canceled)
21 . A system for detecting lung abnormalities, the system comprising:
an image pre-processing module that standardizes an image of an anatomical feature;
a plurality of neural networks in communication with the image pre-processing module and comprising a first neural network that analyzes the X-ray image and optionally one or more additional neural networks that analyzes the image, wherein each neural network is selected from a an image classification and objection detection technologies selected from the group consisting of Faster R-CNN, Inception-Resnet, DenseNet and NasNet network; and
an ensemble classifier that reports presence of an abnormality and its location in the X-ray image based on outputs from the plurality of neural networks.
22 . The system of claim 21 , further comprising a rejection mode that is operable to reject an image in real time based upon pre-programmed criteria.
23 . The system of claim 21 , wherein the image-processing module resizes the chest X-ray image.
24 . The system of claim 23 , further comprising an engineering module that identifies features in the segmented image.
25 . The system of claim 24 , wherein the features are inputs to the ensemble classifier.
26 . The system of claim 25 , wherein at least some of the features represent abnormalities.
27 . The system of claim 21 , further comprising a routine that assigns a confidence score to the features based on a library of known abnormalities.
28 . The system of claim 21 , wherein the ensemble classifier combines outputs of the plurality of neural networks.
29 . The system of claim 21 , wherein at least one of the plurality of neural networks comprises a region proposal network.
30 . The system of claim 21 , wherein the ensemble classifier is a machine learning module trained using chest X-rays, lung CT scans, or lung PET-CT scans.
31 . The system of claim 30 , wherein the ensemble classifier is further trained on clinical outcomes.
32 . A system for detecting abnormalities in tissue, the system comprising:
a computer system comprising non-transitory memory and an operating system comprising a machine learning routine;
a plurality of CT scans from individuals with associated clinical outcomes to enable the machine learning routine to predict clinical outcome based on features in the CT scans;
a classifier operable to compare the CT scans to input data obtained from a patient;
wherein the classifier reports a clinical outcome based on the comparison.
33 . The system of claim 32 , wherein the CT scans represent liver tissue, lung tissue, or brain tissue.
34 . The system of claim 32 , wherein the plurality of CT scans are standardized prior to input to the machine learning routine.