Evidence record · UDINUS · SDGs
- Unit
- Rektorat UDINUS
- Period
- 2025
- Goals
- Record
- 9424cbe8-d5c9-439e-b5f0-80b25a65126c
- Published
- 2026-10-30
Universitas Dian Nuswantoro, through the participation of Ramadhan, Aditya Wahyu; Andono, Pulung Nurtantio; and Soeleman, Moch Arief, conducted research on automated classification of diabetic foot ulcers using the Wagner grading scheme. The activity addressed the limitations of conventional evaluation approaches that rely heavily on clinicians’ visual judgment, which may be subjective and may produce inconsistent assessments. The research proposed a multiclass classification system covering Wagner Grades 0–5 and employed a ResNet-50 deep learning architecture enhanced through transfer learning.
The study used 1,415 clinical wound images that had been labeled and verified by medical experts. The dataset was imbalanced: 543 samples were categorized as Grade 0, 110 as Grade 1, 252 as Grade 2, 145 as Grade 3, 293 as Grade 4, and 72 as Grade 5. To address this imbalance, the research applied random oversampling together with image normalization and data augmentation. These methods were used to increase the diversity of training data and to support recognition of underrepresented Wagner grades. The model also incorporated the Adam optimization algorithm.
Experimental evaluation assessed accuracy, precision, recall, and F1-score. The reported precision was 0.96, recall was 0.95, and F1-score was 0.95. The optimal configuration integrated ResNet-50 with random oversampling and improved prediction accuracy for underrepresented classes, particularly Grades 1 and 5. The classifier demonstrated stable recognition of rare categories without signs of overfitting. The findings indicate that the proposed framework enabled accurate automated classification of diabetic foot ulcers and had potential to assist objective clinical decision-making and accelerate diagnostic processes, especially in resource-constrained healthcare settings.
This UDINUS research relates to RAW_GOALS 3 through its focus on health research addressing diabetic foot ulcer classification and the use of a computational approach to support clinical decision-making.