Related Experiment Video
Updated: Jul 24, 2026
07:15
Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
Published on: August 16, 2020
6.8K
Classification of Diabetic Foot Ulcers from Images Using Machine Learning Approach
Nouf Almufadi1, Haifa F Alhasson1
1Department of Information Technology, College of Computer, Qassim University, Buraydah 51452, Saudi Arabia.
Diagnostics (Basel, Switzerland)
|August 29, 2024
Summary
Diabetic foot ulcers (DFUs) are classified using advanced AI. Convolutional Neural Network (CNN) models achieved high accuracy in distinguishing ischemia (97%) and infection (93%) in DFUs.
Area of Science:
- Medical imaging analysis
- Artificial intelligence in healthcare
- Diabetic complications research
Background:
- Diabetic foot ulcers (DFUs) affect one-third of individuals with diabetes, posing risks of severe complications like amputation.
- Accurate classification of DFU states (ischemia vs. infection) is critical for effective management and preventing misdiagnosis.
- Convolutional Neural Network (CNN) models offer potential for automated DFU state classification.
Purpose of the Study:
- To develop and evaluate an effective classification system for Diabetic Foot Ulcers (DFUs) using various CNN models.
- To compare the performance of different CNN architectures and machine learning classifiers for DFU state classification.
- To identify the optimal model combination for accurate differentiation between ischemic and infected DFUs.
Main Methods:
- Utilized pre-trained CNN models (EfficientNetB0, DenseNet121, ResNet101, VGG16, InceptionV3, MobileNetV2, InceptionResNetV2) with transfer learning.
- Implemented a novel 'head model' for enhanced decision-making within the CNN architectures.
- Evaluated model performance using hyperparameter tuning and integration with machine learning classifiers (Logistic Regression, AdaBoost).
Main Results:
- The integration of the proposed head model with EfficientNetB0 and a logistic regression classifier achieved 97% accuracy in classifying ischemia in DFUs.
- Using the EfficientNetB0 model with the proposed modifications and the AdaBoost classifier resulted in 93% accuracy for classifying infections in DFUs.
- The study demonstrated the effectiveness of CNNs and machine learning classifiers in accurately categorizing DFU states.
Conclusions:
- The developed CNN-based system, particularly EfficientNetB0 with a logistic regression classifier, shows high potential for accurate DFU ischemia classification.
- The EfficientNetB0 model combined with AdaBoost classifier provides a reliable method for DFU infection classification.
- This AI-driven approach can aid in early diagnosis and improve management strategies for Diabetic Foot Ulcers.
