Home // International Journal On Advances in Life Sciences, volume 15, numbers 1 and 2, 2023 // View article
Authors:
Huseyin A. Erdem
Isil Erdem
Semih Utku
Keywords: near-infrared light; chronic venous disorder; deep learning; YOLOv3; indirect augmented reality.
Abstract:
Today's healthcare industry agrees that early diagnosis is just as important as the treatment of diseases. Academic and commercial studies carried out in this direction within the scope of early diagnosis contribute to a better quality of human life. Thanks to devices that offer early diagnosis, treatments can be started when the diseases are at their initial levels, and thus treatment costs can be reduced. However, most of these devices work with harmful rays and are generally used in hospital environments only by specialized staff. In this study, a low-cost early diagnosis system for home use, developed as part of a doctoral thesis, is introduced. The most important aspect of the proposed system that supports the convenience of home use is that it provides a harmless imaging with near-infrared light for the body. The system can detect both Class-1 (spider/telangiectasias vein) and Class-2 (varicose vein) types of the clinical classification of Chronic Venous Disorder, with 4 different classes (in the form of two separate levels as beginner and advanced). In the system, which will monitor the development of the disorders in the superficial veins, the confidence values and positions of the detections in the images were determined by the You Only Look Once version-3 object detection algorithm used in deep learning applications. Confidence values of 0.90 and above were achieved in the object detection experiments performed with Class-1 and Class-2 type artificial patterns. According to the test results obtained, the system was able to detect Chronic Venous Disorder patterns with the values of Accuracy Rate (1), Misclassification Rate (0), Precision (1), Prevalence (0.5) and F-Score (1). The confidence values and positions of the patterns detected in the study are presented to the user/physician with the help of indirect augmented reality visuals as an e-health application that will support a long-term monitoring system. In this way, the beginner and the advanced levels of venous disorder can be monitored by before and after video visuals.
Pages: 20 to 32
Copyright: Copyright (c) to authors, 2023. Used with permission.
Publication date: June 30, 2023
Published in: journal
ISSN: 1942-2660