Technical Description
This invention utilizes a hybrid machine learning approach that combines "topdown" expert eye-tracking patterns with "bottom-up" image saliency maps. By analysing the fixation differences between experts and non-experts, the system precisely identifies and extracts regions of interest (ROI) and pathological features in complex images.
Problems Addressed
- Manual Analysis Inefficiency
- Diagnostic Accuracy Lacks
- High Distractor Interference
- Segmenting Complexity Issues
- Inconsistent Feature Identification
- Search Task Latency
Tech Features
- Expert-Gaze Data Recording
- Attractor-Distractor Region Clustering
- Machine Learning Feature Ranking
- Hybrid Top-Down Knowledge Base
- Enhanced Multi-Saliency Mapping
- Fuzzy Logic Classification
Target Audience
- Medical Diagnostic Software Developers
- Ophthalmology & Radiology Clinics
- Ai Training Data Providers
- Computer Vision Research Labs
- Industrial Quality Control Firms
- Biometric Sensor Manufacturers
Tech ID: P02-1972 TRL 4 Patent Status: Granted Available For Exclusive and Non-exclusive License
