Retinal Lesion Detection Using Optimal Filters Against Confounders
Solution Overview
Problem
Existing automated detection methods for retinal lesions, such as microaneurysms and drusen, struggle to differentiate between target lesions and confounding structures like retinal blood vessels, exudates, and other similar-looking abnormalities, leading to false positives or false negatives, and lack efficient, instantaneous detection capabilities.
Innovation Solution
An optimal filter framework is developed to generate filters that differentiate between target lesions and confounders by using a combination of expert-driven and data-driven approaches, employing mathematical modeling and annotated image samples to create a feature space for rapid lesion detection, allowing for instantaneous feedback.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a more sensitive detection setting is used to detect all potential lesions, then detection sensitivity is improved, but false positives increase due to misidentification of retinal blood vessels and other normal structures
Solution Approach 1:
The patent segments the retinal image into multiple feature spaces (color, texture, shape, size) and processes each dimension separately. By dividing the complex detection task into independent feature extraction and analysis steps, the system can identify lesions based on multiple criteria simultaneously, reducing false positives while maintaining sensitivity.
Solution Approach 2:
The patent transforms 2D retinal images into multiple feature dimensions including color distribution, texture patterns, shape characteristics, and size measurements. This dimensional transformation allows the system to distinguish between lesions and normal structures by analyzing patterns across multiple dimensions rather than relying on a single intensity threshold.
2Reliability
If a specific detection setting is used to reduce false positives, then false positive rate is improved, but detection sensitivity decreases and lesions connected to or close to vasculature are missed
Solution Approach 1:
The patent separates the detection of different lesion types and their spatial relationships from the detection of normal vasculature. By analyzing color, texture, shape, and size features independently, the system can identify lesions near vessels without confusing them with the vessels themselves, maintaining both specificity and sensitivity.
Solution Approach 2:
The patent creates a composite detection approach that combines multiple feature analysis (color filtering, texture analysis, shape recognition, size measurement) into a unified detection system. This composite approach allows the system to maintain high specificity while detecting all lesion types including those adjacent to or connected with vasculature.
3Productivity
If conventional automated detection algorithms are used, then processing speed is improved, but detection accuracy is insufficient compared to human experts
Solution Approach 1:
The patent performs preliminary feature extraction and characterization of retinal structures before the actual detection process. By pre-processing images to identify and segment different feature types (color regions, texture patterns, shape elements, size categories), the system prepares the data in advance, enabling fast and accurate detection without sacrificing precision for speed.
4Ease of operation
If simple detection methods are used, then ease of operation is improved, but the ability to differentiate lesions from confounders deteriorates
Solution Approach 1:
The patent divides the detection system into separate functional modules for color analysis, texture analysis, shape recognition, and size measurement. Each module operates independently and can be understood or adjusted separately, maintaining operational simplicity while achieving sophisticated lesion differentiation through the combination of these segmented functions.
Data Source
AI summary
A method of identifying an object of interest can comprise obtaining first samples of an intensity distribution of one or more object of interest, obtaining second samples of an intensity distribution of confounder objects, transforming the first and second samples into an appropriate first space, performing dimension reduction on the transformed first and second samples, whereby the dimension reduction of the transformed first and second samples generates an object detector, transforming one or more of the digital images into the first space, performing dimension reduction on the transformed digital images, whereby the dimension reduction of the transformed digital images generates one or more reduced images, classifying one or more pixels of the one or more reduced images based on a comparison with the object detector, and identifying one or more objects of interest from the classified pixels.