Automatic Polypoidal Choroidal Vasculopathy Detection in Indocyanine Green Angiography

• Main Authors: Du, Shuozhao, 杜碩釗
• Other Authors: Lin, Weiyang
• Format: Others
• Language: zh-TW
• Published: 2012
• Online Access: http://ndltd.ncl.edu.tw/handle/04756072366589284363

碩士 === 國立中正大學 === 資訊工程研究所 === 100 === With the advancement of ophthalmic technology in recent years, Polypoidal Choroidal Vasculopathy (PCV) has been reported as one of the common causes of blindness in Asian. The location of PCV is usually around macular location and it is characterized by the polypoidal vascular protrusion at the choroidal vessels. In addition, PCV is often associated with subretinal hemorrhage and Pigment Epithelial Detachment (PED). IndoCyanine Green Angiography (ICGA) is considered as the gold standard method in diagnosing PCV currently. But, the procedure is rather time-consuming and could making it inconvenient for physicians to use. For this reason, we develop a computer-aided system which assists the ophthalmologist in locating PCV and provides treatment information in management. Our approach combines "DBICP" registration algorithm, and "Support Vector Machine" machine learning algorithm. We use the curve of intensity changing over time, texture, shape, etc., to charactering PCV in ICGA sequence.The distribution probability map of PCV in each case would provide the physician a useful diagnostic reference. We have conducted several experiments using ICGA images from EVEREST study. The results are promising as compared to the baseline algorithm.