Detection of MA Based on Iris Blood Vessel Segmentation and Classification Using Convolutional Neural Networks (ConvNets)

Abstract

Deep learning and artificial intelligence (AI) play a vital role in the biomedical field for segmenting and classifying various diseases. By the use of AI, highly precise and efficient systems can be developed with which doctors can identify and diagnose diseases at an early stage and without the extensive resources available in specialist clinics. The detection of MA in fundus images remains an open problem in the medical imaging process due to the poor reliability (with existing detection or deduction methods). Detection of diabetic retinopathy in earlier stages is essential for preventing blindness. Detection of microaneurysms (MA) is the initial stage in DR, which is present in the retina with a slight swelling on both sides of the blood vessel. The detection of MA in fundus images remains an open problem in the medical imaging process due to poor reliability. Convolutional neural networks (ConvNets) can achieve an accuracy of 98.358% in the detection of the microaneurysms using publicly available Kaggle datasets. This paper tends to list the various strategies and methods used to detect microaneurysms using ConvNets.