[A novel method of optic disk segmentation based on visual saliency and rotary scanning]

Area of Science:

• Ophthalmology
• Medical Imaging
• Computer Vision

Context:

• Accurate optic disk segmentation is crucial for computer-aided diagnosis of eye diseases.
• Existing methods may lack speed or robustness in segmenting the optic disk from fundus images.

Purpose:

• To propose a novel, rapid method for optic disk localization and boundary segmentation in fundus images.
• To leverage human visual characteristics for enhanced optic disk segmentation.

Summary:

• The method analyzes optic disk features to quickly locate the region of interest.
• It utilizes intensity, color, and spatial distribution to create a saliency map, followed by adaptive thresholding for segmentation.
• A rotary scanning technique is employed to refine the optic disk boundary and minimize vascular interference.

Impact:

• The proposed method demonstrates good performance on the Drishti-GS database, proving to be simple and rapid.
• It meets the standards of eye specialists, suggesting potential for clinical application in computer-aided eye disease diagnosis.