Skip to content
SDGs Center UDINUS

Evidence record · UDINUS · SDGs

Performance Evaluation of YOLOv8 Models for Multi-Class Skin Lesion Detection from Dermoscopic Images

Rektorat UDINUS 2025 Published 2026-10-30

Evidence record · UDINUS · SDGs

Unit
Rektorat UDINUS
Period
2025
Goals
Record
ce82c43a-e43c-4d5d-b466-a3545f9c1695
Published
2026-10-30

Universitas Dian Nuswantoro conducted an activity examining the application of YOLOv8, a convolutional neural network model, for skin cancer detection using the HAM10000 dataset. The activity involved Paramita, Cinantya; Supriyanto, Catur; Šolić, Petar; Wada, Chikamune; and Dzaky, Azmi Abiyyu as participants. The work addressed skin cancer as a significant global health issue and considered the importance of early detection, particularly for melanoma, which poses a high mortality risk due to metastasis.

In the activity, the HAM10000 dataset underwent augmentation. This method was used to address class imbalance and improve detection accuracy. The evaluation covered YOLOv8 variants for the detection of benign and malignant skin cancer lesions. Among the evaluated variants, YOLOv8s demonstrated the highest reported performance, with an mAP50 of 0.890, precision of 0.840, and recall of 0.857. For malignant lesion detection, YOLOv8m showed optimal sensitivity with a recall of 0.857.

The findings indicate the potential of YOLOv8 to enhance diagnostic accuracy for both benign and malignant skin cancer and to contribute to early detection efforts and deep learning applications in healthcare. In relation to the stated goal, the activity provides evidence of health-focused research through the evaluation of a deep learning model for skin lesion detection. This activity represents a research contribution by Universitas Dian Nuswantoro to the application of computational methods in healthcare and to the development of approaches supporting skin cancer detection.

Reference https://www.scopus.com/pages/publications/105012729213?origin=resultslist