Artificial Intelligence Tool Accurately Identifies Pressure Injuries in Clinics
Researchers from Koc University in Istanbul, Turkiye, have designed an artificial intelligence (AI) tool that can more accurately and objectively identify different stages of pressure injuries (PIs). The study used a combination of machine learning and computer vision to classify PI images by stage, with promising results.
Key Takeaways:
- The AI tool was trained on a dataset of 1,691 images, including publicly available images and a private dataset from the Koc University Wound Research Laboratory.
- The tool achieved an average accuracy of 76.92% on the 4-class classification task, with a precision of 87.35% for Stage 1 and 64.72% for Stage 3.
- Grad-CAM was applied to visualize attention areas for further evaluation of the model results.
- The study demonstrated the effectiveness of using AI and computer vision to classify PI images by stage, with potential applications in clinical settings.
- The research was conducted by a team of researchers from Koc University, including Ayse Silanur Demir Uctepe, Ayise Karadag, Ahmet Emin Battal, Cevat Gulec, Eren Ergun, Ahmet Bakcaci, and Cigdem Gunduz Demir Uctepe.
Statistics:
- 1,691 images were used to train and test the AI tool, including 1091 images from the Pressure Injury Image Dataset and 572 images from the private dataset.
- The AI tool achieved an average accuracy of 76.92% on the 4-class classification task.
- The tool demonstrated a precision of 87.35% for Stage 1 and 64.72% for Stage 3.
- The study used a combination of ResNet18, ResNet18-Transformer Encoder Hybrid Model, and DenseNet-121 architectures for training and testing.
Sources:
- NewsRx. Researchers from Koc University Report Findings in Artificial Intelligence (Artificial Intelligence-enabled Staging Classification of Pressure Injuries). Journal of Engineering. October 20, 2025; p 3494.
- Artificial Intelligence-enabled Staging Classification of Pressure Injuries. Advances in Skin & Wound Care, 2025;38(9):480-486.