Treffer: Automated Classification of Diabetic Ulcers and Ordinary Wounds using YOLOv7 for Medical Diagnosis.

Introduction: Diabetic ulcers are a major global health concern, contributing to over 80% of lower limb amputations. Nearly one in four diabetic patients develop these wounds, leading to prolonged hospital stays and increased mortality rates. Current diagnostic methods rely on manual inspection, which is slow, subjective, and error-prone. The recent advancements in deep learning and computer vision have paved the way for automated ulcer classification and diagnosis. While models like Convolutional Neural Networks (CNNs), Faster R-CNN, and Efficient Net have been explored, they exhibit limitations such as high computational costs, false positives, and slow inference times. YOLOv7 offers a promising alternative due to its speed and accuracy in object detection. Aim: To develop an automated classification system for distinguishing between diabetic ulcers and non-ulcer wounds using YOLOv7. The objective is to improve diagnostic accuracy, reduce misdiagnosis rates, and facilitate early intervention for diabetic patients. Materials and Methods: A dataset comprising 1,140 labeled images, including both ulcer and non-ulcer wounds, was used. Preprocessing techniques such as image augmentation, contrast enhancement, and resizing were applied. The dataset was split into 80% for training and 20% for validation. YOLOv7 was trained using multiple hyperparameter tuning strategies, and its performance was evaluated against Faster R-CNN and EfficientNet using accuracy, precision, recall, F1-score, and mean Average Precision (mAP). Results: YOLOv7 achieved an accuracy of 92.5%, outperforming Faster R-CNN (88.7%) and EfficientNet (89.4%). The model effectively detected ulcer boundaries, severity levels, and key features. Compared to traditional methods, the rate of misdiagnosis was reduced by 30%. Conclusion: The present study demonstrates that YOLOv7 can significantly enhance diabetic ulcer classification by improving detection speed and accuracy. The model’s application in medical imaging can assist clinicians in early ulcer diagnosis and treatment planning. Future research will explore integrating thermal and infrared imaging, as well as deploying the model on mobile AI platforms for real-time healthcare applications. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Clinical & Diagnostic Research is the property of JCDR Research & Publications Private Limited and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)