Melanoma Skin Cancer Detection Using EfficientNet and Channel Attention Module
Authors:
- 12 September 2021
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 227-232
Abstract
Melanoma is the most dangerous type of skin cancer and hence early detection is crucial for survival. Classifying melanoma moles from non-melanoma moles is a challenging task since they are visually similar. In this study, we propose a deep transfer learning based approach for melanoma detection, which can able to detect from raw images without any pre-processing. The EfficientNet pre-trained model is utilized to transfer the learned knowledge with the fewer number of skin image samples and to achieve high detection speed. In addition, a channel attention module is included to boost the melanoma-specific features in the classification. Proposed approach is evaluated on UMGC and HAM10000 benchmark datasets with clinical and dermoscopic images, respectively. Experimental results demonstrate proposed work outperforms the state-of-the-art approaches with the classification accuracy of 84.12% and 96.32%, respectively.Keywords
This publication has 24 references indexed in Scilit:
- Knowledge, attitudes and skills in melanoma diagnosis among doctors: a cross sectional study from Sri LankaBMC Research Notes, 2018
- The melanoma skin cancer detection and classification using support vector machine2017
- Adaptable pattern recognition system for discriminating Melanocytic Nevi from Malignant Melanomas using plain photography images from different image databasesInternational Journal of Medical Informatics, 2017
- Set of descriptors for skin cancer diagnosis using non-dermoscopic color images2016
- A Multiview Joint Sparse Representation with Discriminative Dictionary for Melanoma Detection2016
- An intelligent decision support system for skin cancer detection from dermoscopic images2016
- Melanoma detection by analysis of clinical images using convolutional neural network2016
- Rethinking the Inception Architecture for Computer Vision2016
- MED-NODE: A computer-assisted melanoma diagnosis system using non-dermoscopic imagesExpert Systems with Applications, 2015
- ImageNet: A large-scale hierarchical image database2009