Skip to content
SDGs Center UDINUS

Evidence record · UDINUS · SDGs

YOLOv10x Model for Accurate First Detection of Skin Diseases from Dermoscopic Objects

Rektorat UDINUS 2025 Published 2026-10-30

Evidence record · UDINUS · SDGs

Unit
Rektorat UDINUS
Period
2025
Goals
Record
bee321c6-0cd4-4f5c-b48f-e2561782ca4a
Published
2026-10-30

Universitas Dian Nuswantoro conducted research evaluating the effectiveness of the YOLOv10 model for identifying skin lesions in dermoscopic images. The activity involved UDINUS researchers Tjahjono, William Trijanto Dwiputra; Paramita, Cinantya; Supriyanto, Catur; Savicevic, Anamarija Jurcev; Rakasiwi, Sindhu; and Haresta, Alif Agsakli, who participated in the research. The study addressed skin cancer, particularly melanoma, as a global health concern associated with increasing incidence and mortality rates. It also considered the importance of early detection and the variability and delays that may arise when traditional diagnostic methods rely on expert interpretation.

The research used the HAM10000 dataset, which contains 10,015 labelled dermoscopic images categorized into seven lesion types. To address class imbalance, the activity applied preprocessing through oversampling and augmentation techniques. These procedures resulted in a balanced dataset of 3,500 images. The YOLOv10x variant was then trained using optimized parameters. The reported results were a precision of 84.2%, recall of 86.3%, mAP50 of 89.2%, and an F1-score of 85.2%.

Extensive testing further validated the model's ability to distinguish malignant and benign lesions accurately. The research reported that YOLOv10 has real-time processing capacity and robust detection accuracy, indicating potential for clinical applications and more rapid and dependable diagnoses. The activity relates to the stated health-focused goal through research using artificial intelligence and medical imaging. The study describes AI as having transformative potential for medical imaging and identifies future work involving model refinement, dataset expansion, and exploration of real-time implementations for broader medical applications. Through this activity, UDINUS documents research on an AI-based approach to skin-lesion detection and its reported performance.

Reference https://www.scopus.com/pages/publications/105012770412?origin=resultslist