Skin Lesion Segmentation Based on Multi-Scale Attention Convolutional Neural Network

Yun Jiang,Simin Cao,Shengxin Tao,Hai Zhang

doi:10.1109/access.2020.3007512

Yun Jiang, Simin Cao + Show 2 more

Open Access

https://doi.org/10.1109/access.2020.3007512

Copy DOI

Abstract

The incidence of skin cancer around the world is increasing year by year. However, early diagnosis and treatment can greatly improve the survival rate of patients. Skin lesion boundary segmentation is essential to accurately locate lesion areas in dermatoscopic images. It is true that accurate segmentation of skin lesions is still challenging dues to problems such as blurred borders, which requires an accurate and automatic skin lesion segmentation method. In this paper, we propose an end-to-end framework which can perform skin lesion segmentation automatically and efficiently, called the CSARM-CNN (Channel & Spatial Attention Residual Module) model. Each CSARM block of the model combines channel attention and spatial attention to form a new attention module to enhance segmentation results. The multi-scale input images are obtained by the spatial pyramid pooling. Finally, a weighted cross-entropy loss function is used at each side of the output layer to sum the total loss of the model. We evaluated in two published standard datasets, ISIC 2017 and PH2, and achieved competitive results in terms of specificity and accuracy, with 99.03% and 99.45% specificity, 94.96% and 95.23% accuracy, respectively.

Highlights

As the largest organ of the human body, skin is usually directly exposed to the air, which lead to skin diseases one of the most common diseases in humans [1]
Inspired by the above semantic segmentation depth model and the latest advances in attention mechanism, we propose a new framework based on deep CNN, Channel & Spatial Attention Residual Module (CSARM) for automatic segmentation of skin lesions in dermoscopy images
The model is based on the CSARM block, which uses a combination of residual learning, channel attention mechanism and spatial attention mechanism to improve the discriminant and representational ability of CNN

Summary

Introduction

As the largest organ of the human body, skin is usually directly exposed to the air, which lead to skin diseases one of the most common diseases in humans [1]. There are 5.4 million new cases of skin cancer every year [2]. As one of the most lethal malignant skin tumors, causes more than 10,000 deaths each year [3]. Melanoma can be cured by simple resection if it can be detected early. Diagnostic survival rate exceeds 95% and late detection is below 20% [4]. The early diagnosis and early treatment of dermatoses is very important

Methods

Results

Conclusion