Deep Learning Approach to Nailfold Capillaroscopy Based Diabetes Mellitus Detection

Diabetes mellitus is a commonly occurring chronic metabolic disorder which has affected almost 400 million people around the world. It can lead to vascular structure alterations and various renal, cardiovascular, neurologicalcomplications claiming several lives. Since diabetes results changes, NailfoldCapillaroscopy(NFC) based approach be employed for detection of diabetes. NFC an inexpensive, non-invasive method involves acquisition images capillaries in nail bed region using USB digital microscope. Qualitative parameters such as tortuosity, hemorrhages, angiogenesis, elongated capillariesand quantitative like length, width mean capillary density are considered detection. About 600 healthy diabetic subjects were collected further data augmentation was performed increase this 1018 dataset. This paper focuses on obtain employdeep learning-based object algorithm localize these loops nailbed differentiate them into five classes namely, normal, wide, elongated, tortuosity hemorrhages. classification prominent significance medical practitioners helps gauging severity progressionof disorder.