Retinal Blood Vessel Segmentation using Random Forest with Gabor and Canny Edge Features

Recent developments in machine learning increases the researcher's interest in processing the medical images in diagnosis. The medical field requires precise diagnosis to detect the disease. This paper proposed a fusion of features that are extracted from canny edge detector and Gabor feature extractors. These features dimension is huge while combining the features of canny edge detector and Gabor extractor. The Principal Component Analysis applied on the extracted features to reduce the dimension to increase the computational speed. The ensemble method Random Forest is applied on the features to classify the vessel's existence in fundus image. The results have been compared with Decision Tree algorithm. The experiments have proved the Random Forest performed better result with 99.86% accuracy and F1 score 0.997.