Macular Region Detection in Fundus Images via Optic Disk Reference
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for detecting and classifying macular region abnormalities in fundus images using deep learning are inaccurate due to limited training data, leading to poor positioning and detection results.
Innovation Solution
A method that detects the macular region using a target detection model, identifies and positions it based on the optic disk region, and performs multi-modal processing to fuse images for improved detection, incorporating techniques like contrast limited adaptive histogram equalization (CLAHE) and hue saturation value (HSV) color space processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If deep learning model is used for macular region positioning, then positioning can be achieved, but accuracy deteriorates due to limited training data
Solution Approach 1:
The patent introduces the optic disk region as an intermediary reference point to locate the macular region. Instead of directly training the model to detect macular regions from limited data, the system uses the easily detectable optic disk as a mediator, then applies geometric relationships to infer macular position, thereby overcoming the data scarcity problem while maintaining positioning accuracy
Solution Approach 2:
The patent transitions from direct 2D macular region detection to a multi-step approach involving optic disk detection and geometric calculation. By adding the dimensional step of calculating macular position relative to the optic disk center, the system achieves more accurate positioning without requiring additional training data for direct macular detection
2Measurement precision
If multi-modal processing is performed on macular image, then detection accuracy is improved, but processing time increases
Solution Approach 1:
The patent segments the image processing into distinct modular steps: optic disk detection, macular region identification based on geometric relationships, and multi-modal processing only on the extracted macular region. This segmentation allows each step to be optimized independently and reduces overall processing time by avoiding unnecessary processing of the entire fundus image
Solution Approach 2:
The patent performs preliminary actions of optic disk detection and macular region identification before applying computationally intensive multi-modal processing. By pre-processing and extracting only the relevant macular region, the system minimizes the amount of data that requires complex multi-modal analysis, thereby reducing total processing time while maintaining high accuracy
Data Source
AI summary
The present disclosure provides a method, device and equipment for identifying and detecting a macular region in a fundus image. The method includes the following steps: reading a current fundus image to be positioned and detected; detecting a macular region in the fundus image using a target detection model; when the macular region in the fundus image is not detected, detecting an optic disk region in the fundus image, and identifying and positioning the macular region based on the detected optic disk region; based on a positioning result of the macular region, extracting a macular image corresponding to the macular region from the fundus image; and performing multi-modal processing on the macular image, fusing images obtained by the multi-modal processing to obtain a fused image, and detecting whether the macular region is qualified or not according to the fused image.