Image retention and stitching for minimal-flash eye disease diagnosis
Inventors
Clarida, Warren James • Amelon, Ryan Earl Rohret • Shah, Abhay • SUTHER, JACOB PATRICK • Niemeijer, Meindert • Abramoff, Michael David
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-11880976-B2
Publication Date
2024-01-23
Expiration Date
Abstract
Systems and methods are provided herein for minimizing retinal exposure to flash during image gathering for diagnosis. In an embodiment, a system captures a plurality of retinal images of different retinal regions. The system determines that a first portion of a first image does not meet a criterion while a second portion of the first image does meet the criterion, identifies a portion of the retina depicted in the first portion that does not meet the criterion, and determines whether the portion of the retina is depicted in a third portion of a second image and whether the third portion meets the criterion. Responsive to determining that the third portion meets the criterion, the system performs the diagnosis. Responsive to determining that the portion of the retina is not depicted in the second image, the system captures an additional image of the retinal region.
Core Innovation
The invention minimizes exposure of a retina to flash during image gathering for diagnosis. It captures images of different retinal regions and assesses whether image portions meet a criterion. When a retinal portion that does not meet the criterion is adequately depicted in another image, the first retinal region is not flashed again; when that portion is not depicted in the other image, an additional image is captured.
The image set is used for diagnosis, including by passing the images to a fully autonomous machine learning model that outputs a likelihood of a disease condition. The disclosure also describes using a composite image formed by stitching image portions or using analyses of separate image portions. Image quality criteria include exposure, blur, and shadows, and the described system includes image capture, assessment, fallback, recapture, and stitching functions.
Claims Coverage
The three independent claims are claim 1, a method; claim 9, a computer program product; and claim 18, a computer program product comprising modules. They cover three inventive features for assessing retinal image portions and deciding whether to flash a retinal region again.
Quality-based retinal image capture and recapture
Claim 1 captures images of first and second retinal regions and determines whether to flash the first region again by assessing whether an inadequate retinal portion is depicted adequately in the second image. If it is adequately depicted, the method avoids another flash and performs diagnosis using the images; if it is not depicted in the second image, the method determines to capture an additional image. Diagnosis includes passing the images to a fully autonomous machine learning model that outputs a likelihood of a disease condition.
Computer program product for quality-based retinal image capture
Claim 9 covers computer program code that captures images of first and second retinal regions, assesses whether a retinal portion associated with an image portion that fails a criterion is depicted in the second image, and determines whether to avoid flashing the first region again or capture an additional image.
Modular computer program product for retinal image assessment
Claim 18 covers a computer program product with modules for capturing retinal images, assessing whether an inadequate retinal portion is depicted in another image and meets the criterion, avoiding another flash when it does, and determining to capture an additional image when the portion is not depicted in the second image.
The claims cover method and computer-program implementations that use image-quality assessment and representation of retinal portions in other images to determine whether another flash is needed. The claimed computer program products include both code-based and modular implementations.
Stated Advantages
Minimizes exposure of the retina to flash during image gathering for diagnosis by avoiding another flash when the relevant retinal portion is adequately represented in another image.
Documented Applications
Retinal image gathering for diagnosis, including diagnosis using a fully autonomous machine learning model that outputs a likelihood of a disease condition.
Interested in licensing this patent?