Abstract:Automatic skin lesion classification in dermoscopic images is a very challenging task due to the huge intraclass variation, the high degree of interclass visual similarity, low contrast between skin lesion and surrounding normal skin, and the existence of extraneous and intrinsic artifacts. However, existing algorithms for skin lesion classification are developed by leveraging convolutional neural networks (CNNs), and the effectiveness of these algorithms is mostly validated for binary classification of skin lesions. In addition, the relatively low diagnostic sensitivity achieved by these studies demonstrates the uncertainty involved in skin lesion classification. In order to overcome these difficulties, a swin transformer model for multiclass skin lesion classification is proposed by taking advantage of both transformer and CNNs that are based on end-to-end mapping and do not require prior knowledge. Furthermore, the problem of class imbalance is addressed through a weighted cross entropy loss. Moreover, key components of the proposed approach are explored in detail in order to ensure efficient and effective learning process with multiclass data in the skin lesion classification. The proposed method is extensively evaluated on International Skin Imaging Collaboration (ISIC) 2019 Skin Lesion Analysis Towards Melanoma Detection Challenge dataset and achieves a sensitivity, specificity, accuracy, and balanced accuracy value of 82.3%documentclass[12pt]{minimal}usepackage{amsmath}usepackage{wasysym}usepackage{amsfonts}usepackage{amssymb}usepackage{amsbsy}usepackage{mathrsfs}usepackage{upgreek}setlength{oddsidemargin}{-69pt}egin{document}$$82.3\%$$end{document}, 97.9%documentclass[12pt]{minimal}usepackage{amsmath}usepackage{wasysym}usepackage{amsfonts}usepackage{amssymb}usepackage{amsbsy}usepackage{mathrsfs}usepackage{upgreek}setlength{oddsidemargin}{-69pt}egin{document}$$97.9\%$$end{document}, 97.2%documentclass[12pt]{minimal}usepackage{amsmath}usepackage{wasysym}usepackage{amsfonts}usepackage{amssymb}usepackage{amsbsy}usepackage{mathrsfs}usepackage{upgreek}setlength{oddsidemargin}{-69pt}egin{document}$$97.2\%$$end{document}, and 82.3%documentclass[12pt]{minimal}usepackage{amsmath}usepackage{wasysym}usepackage{amsfonts}usepackage{amssymb}usepackage{amsbsy}usepackage{mathrsfs}usepackage{upgreek}setlength{oddsidemargin}{-69pt}egin{document}$$82.3\%$$end{document}, respectively. Experimental results demonstrate that the proposed method has the highest balanced accuracy value and outperforms most of the other state-of-the-art methods in multiclass skin lesion classification.

Enhancing Skin Cancer Diagnosis Using Swin Transformer with Hybrid Shifted Window-Based Multi-head Self-attention and SwiGLU-Based MLP

Addressing Challenges in Skin Cancer Diagnosis: A Convolutional Swin Transformer Approach

A novel Swin transformer approach utilizing residual multi-layer perceptron for diagnosing brain tumors in MRI images

Enhancing Melanoma Diagnosis with Advanced Deep Learning Models Focusing on Vision Transformer, Swin Transformer, and ConvNeXt

A Novel Vision Transformer Model for Skin Cancer Classification

A novel deep learning framework based swin transformer for dermal cancer cell classification

Skin Cancer Segmentation and Classification Using Vision Transformer for Automatic Analysis in Dermatoscopy-Based Noninvasive Digital System

Towards improved fundus disease detection using Swin Transformers

Skin Cancer Detection utilizing Deep Learning: Classification of Skin Lesion Images using a Vision Transformer

Predicting microsatellite instability and key biomarkers in colorectal cancer from H&E-stained images: achieving state-of-the-art predictive performance with fewer data using Swin Transformer

A Deep-Ensemble-Learning-Based Approach for Skin Cancer Diagnosis

Early Cervical Cancer Diagnosis with SWIN-Transformer and Convolutional Neural Networks

Transforming Skin Cancer Diagnosis: A Deep Learning Approach with the Ham10000 Dataset

Multi-Level Swin Transformer Enabled Automatic Segmentation and Classification of Breast Metastases

An improved transformer network for skin cancer classification

Advancing brain tumor detection: harnessing the Swin Transformer’s power for accurate classification and performance analysis

An efficient multi-class classification of skin cancer using optimized vision transformer

Swin transformers are robust to distribution and concept drift in endoscopy-based longitudinal rectal cancer assessment

SwinCNN: An Integrated Swin Transformer and CNN for Improved Breast Cancer Grade Classification

Predicting microsatellite instability and key biomarkers in colorectal cancer from H&E-stained images: Achieving SOTA predictive performance with fewer data using Swin Transformer

Multiclass skin lesion classification in dermoscopic images using swin transformer model