A Transfer Learning Approach for Detection and Classification of Skin CancerPages 205-214
Abstract:
Skin disease is one of the major problems in the world. Skin cancer, infection, acne, psoriasis, and eczema can significantly affect the health of a person. Skin cancers such as melanoma, squamous cell carcinoma, and basal cell carcinoma are the most concerning problems that require early attention to save lives and simplify the treatment. Other conditions like eczema and psoriasis, though not life-threatening, impact mental health and appearance, underscoring the need for timely intervention. Machine Learning and Deep Learning have become a transformative tool in dermatology, leveraging deep learning and computer vision to diagnose skin diseases effectively. In this paper, we presented a model that can classify between Melanoma and Benign types of skin cancer using EfficientNetB7 the Transfer Learning Model with a good test accuracy of 84.09% on the ISIC dataset. The proposed model achieves mean precision, recall, and F1-score is 96.11%, 79.15%, and 86.63% respectively.
Keywords: Machine Learning, Deep learning, Transfer Learning, Skin Lesion, EfficientNet Model
|
PDF 