Improving the Accuracy of Diagnostic Imaging using Artificial Intelligence: A Method for Assessing Necrotic Tissue in
Yuka Kimura1, Kento Ikuta1, Makoto Ohga1
1Department of Plastic and Reconstructive Surgery, School of Medicine, Faculty of Medicine, Tottori University, Yonago 683-8504, Japan.
Yonago Acta Medica
|August 14, 2025
Summary
Artificial intelligence (AI) improves pressure injury assessment. A binary classification model (BCM) demonstrated superior accuracy in identifying necrotic tissue compared to a categorical model (CCM).
Area of Science:
- Medical Informatics
- Artificial Intelligence in Healthcare
- Wound Care Technology
Background:
- Accurate pressure injury assessment, particularly necrotic tissue evaluation using the DESIGN-R® scale, is crucial in clinical practice.
- The study addresses the need for enhanced diagnostic consistency and accuracy in pressure injury evaluation.
- Integration of Artificial Intelligence (AI) is explored to improve the objective assessment of necrotic tissue.
Purpose of the Study:
- To develop and evaluate AI models for classifying necrotic tissue in pressure injuries.
- To compare the performance of a Categorical Classification Model (CCM) and a Binary Classification Model (BCM) for necrotic tissue assessment.
- To determine the potential of AI in standardizing pressure injury evaluation based on the DESIGN-R® scale.
Main Methods:
- A retrospective observational study utilized electronic medical records and wound photographs (2014-2022).
- Two supervised deep learning models, a CCM and a BCM, were developed for necrotic tissue classification (n0, N3, N6).
- Model performance was assessed using standard classification metrics including recall, accuracy, precision, and F-1 score.
Main Results:
- The Binary Classification Model (BCM) achieved higher recall rates across all necrosis categories (n0: 0.9074, N3: 0.9884, N6: 1.0000) compared to the Categorical Classification Model (CCM).
- The BCM demonstrated superior overall performance with an accuracy of 0.8711, precision of 0.8418, and F-1 score of 0.8508.
- The BCM consistently outperformed the CCM across all evaluated performance indicators.
Conclusions:
- AI, specifically the binary classification approach, significantly enhances the assessment of necrotic tissue in pressure injuries.
- The BCM shows strong potential as a reliable tool for assisting clinicians in objective and standardized pressure injury evaluation.
- This AI-driven method supports the consistent application of the DESIGN-R® framework for pressure injury management.
