TransSLC: skin lesion classification in dermatoscopic images using transformers.

Sarker, Md Mostafa Kamal; Moreno-García, Carlos Francisco; Ren, Jinchang; Elyan, Eyad

Abstract

Early diagnosis and treatment of skin cancer can reduce patients' fatality rates significantly. In the area of computer-aided diagnosis (CAD), the Convolutional Neural Network (CNN) has been widely used for image classification, segmentation, and recognition. However, the accurate classification of skin lesions using CNN-based models is still challenging, given the inconsistent shape of lesion areas (leading to intra-class variance) and inter-class similarities. In addition, CNN-based models with massive downsampling operations often result in loss of local feature attributes from the dermatoscopic images. Recently, transformer-based models have been able to tackle this problem by exploiting both local and global characteristics, employing self-attention processes, and learning expressive long-range representations. Motivated by the superior performance of these methods, in this paper we present a transformer-based model for skin lesion classification. We apply a transformers-based model using bidirectional encoder representation from the dermatoscopic image to perform the classification task. Extensive experiments were carried out using the public dataset HAM10000, and promising results of 90.22%, 99.54%, 94.05%, and 96.28% in accuracy, precision, recall, and F1 score respectively, were achieved. This opens new research directions towards further exploration of transformers-based methods to solve some of the key challenging problems in medical image classification, namely generalisation to samples from a different distribution.

Citation

SARKER, M.M.K., MORENO-GARCÍA, C.F., REN, J. and ELYAN, E. 2022. TransSLC: skin lesion classification in dermatoscopic images using transformers. In Yang, G., Aviles-Rivero, A., Roberts, M. and Schönlieb, C.-B. (eds.) Medical image understanding and analysis: proceedings of 26th Medical image understanding and analysis 2022 (MIUA 2022), 27-29 July 2022, Cambridge, UK. Lecture notes in computer sciences, 13413. Cham: Springer [online], pages 651-660. Available from: https://doi.org/10.1007/978-3-031-12053-4_48

• Presentation Conference Type: Conference Paper (published)
• Conference Name: 26th Medical image understanding and analysis 2022 (MIUA 2022)
• Start Date: Jul 27, 2022
• End Date: Jul 29, 2022
• Acceptance Date: Jun 1, 2022
• Online Publication Date: Jul 25, 2022
• Publication Date: Jul 25, 2022
• Deposit Date: Jul 28, 2022
• Publicly Available Date: Jul 26, 2023
• Publisher: Springer
• Peer Reviewed: Peer Reviewed
• Pages: 651-660
• Series Title: Lecture notes in computer science
• Series Number: 13413
• Series ISSN: 0302-9743; 1611-3349
• Book Title: Medical image understanding and analysis
• ISBN: 9783031120527
• DOI: https://doi.org/10.1007/978-3-031-12053-4_48
• Keywords: Computer aided diagnosis; Skin lesion classification; Deep learning; Convolutional neural networks; Transformers
• Public URL: https://rgu-repository.worktribe.com/output/1721434