Training a neural network for a predictive aortic aneurysm detection system
AssigneeRowan University
InventorsYupeng Li, Hieu Duc Nguyen, Shao Tang
Life of this patent
Estimating the term…
Abstract
Systems and methods for detecting aortic aneurysms using ensemble based deep learning techniques that utilize numerous computed tomography (CT) scans collected from numerous de-identified patients in a database. The system includes software that automates the analysis of a series of CT scans as input (in DICOM file format) and provides output in two dimensions: (1) ranking CT scans by risks of adverse events from aortic aneurysm, (2) providing aortic aneurysm size estimates. A repository of CT scans may be used for training of deep neural networks and additional data may be drawn from localized patient information from institutions and hospitals which grant permission.
Claims20
A computer-implemented method for detecting aortic aneurysms, comprising:
- receiving a series of images associated with a thoracic region of a patient, the series of images comprising scans of at least one aortic region;
- iteratively enhancing each image in the series of images to detect and isolate images of respective aortic regions;
- for each enhanced image, determining a center of mass point in an x-y plane corresponding to portions of each image corresponding to an aortic region;
- constructing a centerline of an aorta portion in response to determining the center of mass for each successive image;
- extrapolating a series of aorta diameters based on a vector of the centerline for each of the series of images aligned according to the center of mass point calculations;
- and
- generating a communication to a diagnostic engine that identifies the maximum extrapolated aorta dimension.
A computer system for detecting aortic aneurysms, comprising:
- a means for receiving a series of images associated with a thoracic region of a patient, the series of images comprising scans of at least one aortic region;
- a means for iteratively enhancing each image in the series of images to detect and isolate images of respective aortic regions;
- for each enhanced image, a means for determining a center of mass point in an x-y plane corresponding to portions of each image corresponding to an aortic region;
- a means for constructing a centerline of an aorta portion in response to determining the center of mass for each successive image;
- a means for extrapolating a series of aorta diameters based on a vector of the centerline for each of the series of images aligned according to the center of mass point calculations;
- and
- a means for generating a communication to a diagnostic engine that identifies the maximum extrapolated aorta dimension.
A computer-implemented method for training a neural network to detect aortic aneurysms, comprising:
- collecting a first data set corresponding to aortic images of multiple patients that are stored in an aortic image database;
- collecting a second data set from one or more remote computer users about aortic diagnostics of one or more patients associated with the one or more remote computer users that are exclusive of the first data set;
- creating a first training set comprising the collected first data set;
- training the neural network in a first stage using the first training set;
- applying one or more transformations to the second data set;
- creating a second training set for a second stage of training by altering the first training set with data from the second data set;
- and
- altering the neural network in a second stage using the second training set.
Dependencies are inferred from the claim text.