CNN Avascular Map Generation from OCTA and Reflectance Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for diagnosing diabetic retinopathy, such as Fluorescein Angiography, are limited by the need for invasive fluorescent dyes, require specialized equipment, and are subjective, making them inaccessible to low-resource environments and time-consuming, while Optical Coherence Tomography Angiography (OCTA) is safer but lacks effective tools for identifying avascular areas.
Innovation Solution
A system utilizing a convolutional neural network (CNN) that processes OCT reflectance and OCTA images to generate avascular maps, overcoming the limitations of traditional methods by providing a label-free, faster, and more accessible means of identifying avascular areas in diabetic retinopathy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If Fluorescein Angiography is used to diagnose diabetic retinopathy, then avascular areas can be assessed, but the procedure requires invasive fluorescent dyes, specialized equipment, and is time-consuming
Solution Approach 1:
The patent uses Optical Coherence Tomography Angiography (OCTA) to create a label-free copy or alternative representation of retinal vascular structures, replacing the need for fluorescent dye-based angiography. OCTA captures vascular information through optical coherence tomography without requiring contrast agents, thereby eliminating the complexity and invasiveness of FA while preserving the ability to assess avascular areas
Solution Approach 2:
The patent replaces the mechanical/chemical system of fluorescent dye administration and FA imaging with an optical system based on OCTA. This substitution eliminates the need for invasive dye injection and specialized FA equipment, using instead optical coherence tomography principles to achieve vascular imaging through label-free methods
2Measurement precision
If Fluorescein Angiography is used to diagnose diabetic retinopathy, then avascular areas can be identified, but the procedure is highly subjective and requires specialized training
Solution Approach 1:
The patent implements automated analysis algorithms that enable the OCTA system to self-analyze retinal images and objectively identify avascular areas without requiring specialized clinician training. The system processes the OCTA data through computational algorithms that automatically quantify avascular regions, eliminating the subjectivity and training requirements associated with manual FA interpretation
Solution Approach 2:
The patent incorporates automated feedback mechanisms where the system provides objective quantitative measurements of avascular areas directly from OCTA images. This feedback loop replaces subjective clinician assessment with algorithm-driven objective analysis, reducing the need for specialized training while maintaining or improving measurement precision
3Measurement precision
If Fluorescein Angiography is used to diagnose diabetic retinopathy, then disease severity can be assessed, but the procedure is inaccessible to low-resource environments
Solution Approach 1:
The patent makes OCTA technology universally applicable by demonstrating its ability to perform multiple functions: it can assess avascular areas, evaluate disease severity, and provide quantitative measurements all through a single label-free imaging modality. This multi-functionality allows OCTA to replace FA across diverse settings, including low-resource environments where FA equipment and expertise are unavailable
Solution Approach 2:
The patent promotes the use of OCTA as a cost-effective alternative to FA, eliminating the need for expensive fluorescent dyes and specialized FA equipment. By using readily available OCT technology with automated analysis, the system provides accessible disease severity assessment that can be deployed in resource-constrained settings without requiring expensive consumables or specialized infrastructure
4Reliability
If Fluorescein Angiography is used to diagnose diabetic retinopathy, then treatment threshold can be determined, but the procedure requires administering dyes that cause complications
Solution Approach 1:
The patent converts the harmful aspect of dye administration in FA into a benefit by using label-free OCTA imaging. Instead of exposing patients to potentially harmful fluorescent dyes that can cause nausea and other complications, the system uses optical coherence tomography to safely capture vascular information without any contrast agents, thereby eliminating patient harm while maintaining the ability to determine treatment thresholds reliably
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The system achieves accurate detection of avascular areas with an accuracy of 81.88%, comparable to high-quality 3x3 mm OCTA scans, without excluding low-quality scans, and provides a non-invasive, cost-effective solution for diagnosing diabetic retinopathy, especially in resource-constrained settings.
Implementation Method 1
generating, using at least one first convolution layer in the multi-scale block, a first output image by cross-correlating at least one first filter with at least one input image using a first dilation rate
Data Source
AI summary
An example method includes generating, using a multi-scale block of a convolutional neural network (CNN), a first output image based on an optical coherence tomography (OCT) reflectance image of a retina and an OCT angiography (OCTA) image of the retina. The method further includes generating, using an encoder of the CNN, at least one second output image based on the first output image and generating, using a decoder of the CNN, a third output image based on the at least one second output image. An avascular map is generated based on the third output image. The avascular map indicates at least one avascular area of the retina depicted in the OCTA image.