Real-Time Ophthalmic Image Artifact Detection With Two-Stage Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Artifacts in image data during cataract surgery, such as illumination glint, motion artifacts, and debris, lead to measurement errors and poor surgical outcomes due to their unnoticed presence, necessitating improved image data processing techniques.
Innovation Solution
A two-stage machine learning model is employed to analyze image data from intraoperative aberrometers, comprising a feature extraction stage and a classification stage, to identify and filter out artifacts, providing real-time feedback to medical practitioners.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional image processing methods are used during cataract surgery, then the surgical procedure can proceed, but measurement errors occur due to undetected artifacts in the image data
Solution Approach 1:
The system performs preliminary artifact detection and classification before refractive measurements are taken. The machine learning model analyzes images for artifacts such as illumination glint, motion artifacts, and debris, and provides feedback to prevent inaccurate measurements from being recorded, thereby ensuring measurement precision and surgical reliability
Solution Approach 2:
The system implements a feedback mechanism where the machine learning model continuously monitors image quality, identifies artifacts, and provides real-time feedback to the surgical system. This feedback loop allows the system to reject or correct images with artifacts, preventing measurement errors and improving both measurement precision and surgical outcome reliability
2Measurement precision
If a machine learning model is implemented to detect artifacts, then measurement accuracy improves, but device complexity increases
Solution Approach 1:
The patent extracts the artifact detection function into a separate machine learning model that operates independently from the main surgical instrumentation. This modular approach allows the complex AI processing to be isolated, with the model receiving images and returning artifact probability scores, thereby improving measurement precision while managing device complexity through functional separation
Solution Approach 2:
The machine learning model serves as an intermediary between the image capture system and the refractive measurement system. It acts as a mediator that processes images, identifies artifacts, and provides quality assessment feedback without directly interfering with the core surgical functions, thus improving measurement precision while containing complexity within the intermediary layer
Data Source
AI summary
Certain aspects of the present disclosure provide a system for processing image data from an intraoperative diagnostic device in real-time during an ophthalmic procedure. The system comprises an image capture element that captures a grayscale image of a first size and an image processing element that scales the grayscale image from the first size to a second size. The system also comprises a two-stage classification model comprising: a feature extraction stage to process the scaled grayscale image and generate a feature vector based on the scaled grayscale image and a classification stage to process the feature vector and generate an output vector. The image processing element is further configured to determine an image quality of the obtained grayscale image based on the output vector for display to an operator and the image quality of the obtained grayscale image indicates a probability that the obtained grayscale image includes an artifact.