OCT Deep Learning for Subject-Specific aVEGF Treatment Scheduling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for determining anti-vascular endothelial growth factor (aVEGF) injection schedules for treating wet age-related macular degeneration are subjective and prone to excessive or insufficient treatments, lacking a subject-specific, objective approach to effectively manage the condition while minimizing injections.
Innovation Solution
An optical coherence tomography (OCT) image is processed using a deep convolutional neural network to identify retinal features, generate cropped images, and predict a treatment schedule that prevents vessel leakage by analyzing patch-specific neural networks and ensemble models, providing a subject-specific treatment frequency and interval recommendation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If aVEGF agents are administered via intravitreal injection to treat wet AMD, then the effectiveness of treating the condition is improved, but the side effects and patient discomfort increase
Solution Approach 1:
The patent changes the parameter of treatment frequency by using deep learning analysis of OCT images to predict the optimal interval between injections for each patient. This allows extending the time between injections (reducing treatment frequency) while maintaining treatment effectiveness, thereby reducing side effects and patient discomfort associated with frequent intravitreal injections.
Solution Approach 2:
The patent replaces the manual clinical decision-making process (mechanical system of doctor assessment) with an automated deep learning system that analyzes OCT images. This substitution enables more precise, objective determination of treatment intervals, optimizing the balance between treatment effectiveness and minimizing unnecessary injections that cause side effects.
2Productivity
If treat-and-extend protocol is used to extend inter-injection intervals, then the reduction in injection frequency is improved, but the risk of new leakage before sufficient injection frequency is reached increases
Solution Approach 1:
The patent performs preliminary analysis of OCT images using deep learning models to predict the optimal treatment interval before the next injection is scheduled. This preliminary action allows clinicians to confidently extend inter-injection intervals with quantified risk assessment, reducing the likelihood of new leakage while minimizing unnecessary injections.
Solution Approach 2:
The system uses deep learning analysis of OCT images to provide feedback on the current retinal status and predict the optimal timing for the next injection. This feedback mechanism enables dynamic adjustment of treatment intervals, extending them when safe and preventing new leakage, while maintaining high productivity by avoiding unnecessary injections.
3Ease of operation
If guess-and-check approaches are used to determine aVEGF injection schedules, then the simplicity of the method is maintained, but the precision of treatment scheduling deteriorates
Solution Approach 1:
The patent replaces the subjective guess-and-check approach with an automated deep learning system that objectively analyzes OCT images to predict optimal treatment intervals. This substitution maintains ease of operation for clinicians (who simply input images and receive predictions) while dramatically improving the precision and accuracy of treatment scheduling through data-driven insights.
Solution Approach 2:
The system creates a digital model (copy) of the patient's retinal structure from OCT images and uses deep learning to analyze this digital representation. This allows precise measurement and prediction of treatment needs without requiring complex manual assessment, maintaining simplicity while improving accuracy.
Data Source
AI summary
Systems and methods relate to processing optical tomography coherence (OCT) images to predict characteristics of a treatment to be administered to effectively treat age-related macular degeneration. The processing can include pre-processing the image by flattening and/or cropping the image and processing the pre-processed image using a neural network. The neural network can include a deep convolutional neural network. An output of the neural network can indicate a predicted frequency and/or interval at which a treatment (e.g., anti-vascular endothelial growth factor therapy) is to be administered so as to prevent leakage of vasculature in the eye.