Classification of Diabetic Foot Ulcers Using Class Knowledge Banks.

Area of Science:

• Medical imaging
• Artificial intelligence
• Diabetology

Background:

• Diabetic foot ulcers (DFUs) are a major diabetes complication.
• Accurate classification of DFU infection and ischemia is crucial for treatment.
• Current deep learning models for DFU classification have limitations in utilizing training data knowledge.

Purpose of the Study:

• To propose a novel deep learning approach for DFU classification.
• To enhance DFU image classification by effectively utilizing knowledge from training data.
• To improve the accuracy of identifying infection and ischemia in diabetic foot ulcers.

Main Methods:

• Development of Class Knowledge Banks (CKBs) composed of trainable units.
• CKBs extract and represent class-specific knowledge from training data.
• Prediction is based on image representation similarity to CKB units, enhancing traditional deep learning.

Main Results:

• The proposed method significantly improves the performance of DFU infection classification.
• The method also enhances the accuracy of DFU ischemia classification.
• Experimental results validate the effectiveness of CKBs in DFU assessment.

Conclusions:

• Class Knowledge Banks offer a promising approach to enhance deep learning models for DFU analysis.
• This method improves the utilization of training data, leading to more accurate clinical assessments.
• The findings contribute to better management and treatment planning for diabetic foot ulcers.