Automatic Segmentation of Polypoidal Choroidal Vasculopathy from Indocyanine Green Angiography Using Spatial and...

Area of Science:

• Ophthalmology
• Medical Imaging
• Computer-Aided Diagnosis

Background:

• Polypoidal choroidal vasculopathy (PCV) is a significant cause of vision impairment.
• Accurate detection and quantification of polypoidal lesions are crucial for PCV management.
• Current diagnostic methods may lack objectivity and quantitative precision.

Purpose of the Study:

• To develop an automated computer-aided diagnostic (CAD) tool for detecting and quantifying polypoidal regions in indocyanine green angiography (ICGA) images.
• To assess the performance of the developed CAD tool in a multi-center study.

Main Methods:

• Utilized ICGA sequences from 59 treatment-naïve PCV patients across five Asian countries (EVEREST study).
• Developed a detection algorithm incorporating both temporal and spatial features for polypoidal lesion characterization.
• Employed leave-one-out cross-validation and a fixed detection threshold for performance evaluation.

Main Results:

• The CAD system achieved a high average accuracy of 0.9126 (sensitivity=0.9125, specificity=0.9127) in detecting polyps.
• Spatial variances were identified as the most discriminative feature for polyp identification.
• Combining spatial and temporal features demonstrated potential for improved detection accuracy.

Conclusions:

• The developed software represents a novel tool for the detection and quantification of polypoidal regions in ICGA images.
• This preliminary study highlights the potential of CAD for objective evaluation and monitoring of PCV.
• The system offers ophthalmologists an intuitive way to visualize polyps and obtain quantitative data for improved patient care.