研究目的
To develop a system for automatic detection and therapeutic monitoring of glaucoma by analyzing OCT-GCL and VF images to calculate percentages of healthy, sick, and dead regions for early diagnosis.
研究成果
The proposed system effectively segments and characterizes OCT-GCL and VF images to detect glaucoma early, with results validated by ophthalmologists. It aids in reliable diagnosis and monitoring, though further enhancements with automated methods and larger datasets are recommended for broader application.
研究不足
The database is local and small (58 OCT and 21 VF images), which may limit generalizability. Segmentation relies on fixed thresholds that might not adapt to all image variations. Future work could involve larger databases and machine learning for improved automation and accuracy.
1：Experimental Design and Method Selection:
The methodology involves preprocessing (ROI extraction, noise reduction), segmentation using thresholding and morphological operators, characterization by calculating region percentages, and classification based on expert-validated thresholds.
2：Sample Selection and Data Sources:
A local database of 58 OCT-GCL images (28 normal, 30 abnormal) and 21 VF images (15 glaucomatous, 6 healthy) from patients, acquired using specific devices.
3：List of Experimental Equipment and Materials:
OCT device from Optovue, VF device OPTOPOL type PTS 1000, images in JPG format.
4：Experimental Procedures and Operational Workflow:
Steps include manual ROI reduction, thresholding for blind task elimination, median filtering, segmentation into regions based on color or gray levels, percentage calculation, and classification using if-then-else logic.
5：Data Analysis Methods:
Calculation of area percentages using pixel sums, comparison with fixed thresholds for classification.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
