A novel approach for skin lesion symmetry classification with a deep learning model

Abstract

Symmetry is a distinguishing feature when diagnosing the malignancy of skin lesions, those with an irregular shape –asymmetry– are more likely to have a worse prognosis. This work presents a novel approach for skin lesion symmetry classification of dermoscopic images based on deep learning techniques. Also, we introduce a new dataset of labels for 615 skin lesions. During experimentation, we also evaluate whether it is beneficial to rely on transfer learning from pre-trained CNNs or traditional learning-based methods. As a result, we present a new simple, robust, and fast classification pipeline that outperforms methods based on traditional approaches or pre-trained networks, with a weighted-average F1-score of 64.5%.