Scholarlink

<p>Diabetic Foot (DF) is a complication of diabetes, a condition that affects the feet as a result of poor circulation and peripheral nerve disorders due to long-term high blood sugar. Failure to detect DF early causes serious problems, such as amputation of the lower limbs if it is not treated initially. To avoid the consequences of this disease, real-time diagnosis of DF is crucial. This paper aims to predict diabetic feet in real time using thermal visualization and Machine Learning (ML) algorithms that detect abnormal changes in skin temperature. Thus, it can avoid serious complications caused by DF. Three ML algorithms, including Support Vector Machine)SVM(, Random Forest)RF(, and K-Nearest Neighbors (KNN), and one deep learning model, Visual Geometry Group 16-layer network (VGG16), implemented in Python to classify and predict healthy/unhealthy feet using feature extraction from foot thermal images. Different thermal imaging data from 112 individuals (95 from Kaggle and 17 from Endocrine Center in Baghdad) were collected, which amounted to 2,240 thermal images, equally divided into healthy and unhealthy. 80% of the dataset was used to train, and 20% for testing the ML algorithm. The algorithms were evaluated regarding several performance metrics, such as accuracy, sensitivity, specificity, precision, F1-Score, and Area Under the Curve (AUC). The KNN algorithm outperformed other ML algorithms used in this research and those used in previous research in terms of accuracy, reaching 99.47% compared to the other algorithms used in this research and those used in previous research. The use of ML algorithms with thermal imaging shows high success in detecting temperature variations and identifying early-stage complications of DF.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

High-Accuracy Diabetic Foot Prediction Using Thermal Imaging and KNN-Based Machine Learning

  • 期刊论文

  • 发表日期：

    2025年11月26日

查看全文
  • Ahmed Akeel Ali,
  • Sadik Kamel Gharghan,
  • Adnan Hussein Ali

摘要

Diabetic Foot (DF) is a complication of diabetes, a condition that affects the feet as a result of poor circulation and peripheral nerve disorders due to long-term high blood sugar. Failure to detect DF early causes serious problems, such as amputation of the lower limbs if it is not treated initially. To avoid the consequences of this disease, real-time diagnosis of DF is crucial. This paper aims to predict diabetic feet in real time using thermal visualization and Machine Learning (ML) algorithms that detect abnormal changes in skin temperature. Thus, it can avoid serious complications caused by DF. Three ML algorithms, including Support Vector Machine)SVM(, Random Forest)RF(, and K-Nearest Neighbors (KNN), and one deep learning model, Visual Geometry Group 16-layer network (VGG16), implemented in Python to classify and predict healthy/unhealthy feet using feature extraction from foot thermal images. Different thermal imaging data from 112 individuals (95 from Kaggle and 17 from Endocrine Center in Baghdad) were collected, which amounted to 2,240 thermal images, equally divided into healthy and unhealthy. 80% of the dataset was used to train, and 20% for testing the ML algorithm. The algorithms were evaluated regarding several performance metrics, such as accuracy, sensitivity, specificity, precision, F1-Score, and Area Under the Curve (AUC). The KNN algorithm outperformed other ML algorithms used in this research and those used in previous research in terms of accuracy, reaching 99.47% compared to the other algorithms used in this research and those used in previous research. The use of ML algorithms with thermal imaging shows high success in detecting temperature variations and identifying early-stage complications of DF.