Diabetic Foot Ulcer Ischemia and Infection Classification Using EfficientNet Deep Learning Models.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Diabetic Foot Ulcers

Background:

• Diabetic Foot Ulcers (DFUs) pose a significant risk for limb amputation due to infection and ischemia.
• Early and accurate detection of these complications is crucial for effective treatment.

Purpose of the Study:

• To develop and evaluate a deep learning-based system for detecting infection and ischemia in DFUs from images.
• To compare the performance of the EfficientNet model against established deep learning architectures.

Main Methods:

• Augmentation of a DFU image dataset using geometric and color transformations.
• Implementation of the EfficientNet deep learning model for binary classification of infection and ischemia.
• Comparison with baseline models including ResNet, Inception, and Ensemble CNN.

Main Results:

• EfficientNet achieved high accuracy: 99% for ischemia and 98% for infection.
• EfficientNet significantly outperformed baseline models (e.g., 87% for ResNet/Inception, 90%/73% for Ensemble CNN).
• EfficientNet processed images 5-10 times faster than baseline models.

Conclusions:

• EfficientNet demonstrates high efficacy and efficiency for DFU infection and ischemia classification.
• This deep learning approach offers a promising tool for clinical decision support in managing DFUs.
• The developed system has the potential to reduce amputation rates by enabling timely interventions.