Implementation of Detectron2 Network for The Diagnosis of Chronic Venous Insufficiency Condition Based On Infrared Thermal Imaging

Abstract

Chronic venous insufficiency (CVI) is primarily caused by venous reflux, which generally occurs due to incompetent vein valves. The symptoms include lower extremity edema, discomfort, and skin changes caused by obstructed or incompetent venous valves and venous hypertension. Aging, obesity, genetic susceptibility, hormonal changes including menopause and pregnancy, sedentary lifestyle, leg injury, smoking, and hypertension are all risk factors for venous insufficiency. It is extremely crucial to detect this disease early in order to avoid unnecessary complications. The efficacy of the treatment depends on the diagnosis of this medical condition. To aid this problem, we use infrared thermal images acquired from the control and study population. The main objective of this research is to detect the dilated and twisted veins for the diagnosis of CVI in lower limbs. In this article, we proposed an automatic detection and instance segmentation method based on R-CNN models of deep learning (DL) using Detectron2. The experimental findings demonstrated that the proposed detection method for CVI using the Detectron2 network detected abnormal veins with an improved precision of 84.4% and recall of 86.7%. The study concludes that Detectron2 with Mask and Faster R-CNN is a reasonable model for detecting the CVI affected leg from the thermal image and classifying whether the image is normal or abnormal. Hence, CVI condition can be efficiently diagnosed using infrared thermography and deep learning.