Skin cancer classification based on CNN model with attention mechanism

Zian Song; Yi Zhou

doi:10.1117/12.2636772

10 June 2022 Skin cancer classification based on CNN model with attention mechanism

Zian Song, Yi Zhou

Proceedings Volume 12179, Second International Conference on Medical Imaging and Additive Manufacturing (ICMIAM 2022); 1217917 (2022) https://doi.org/10.1117/12.2636772
Event: Second International Conference on Medical Imaging and Additive Manufacturing (ICMIAM 2022), 2022, Xiamen, China

Abstract

Early diagnosis of skin cancer plays an important role in cure rate increasement, and deep learning models adopting neural networks for malignant & benign skin mole image classification obtained great accuracy. However, most of the existing works used the skin data in a brute force way that pays equal attention to both the skin lesion and irrelevant features that compromise overall accuracy and bring severe bias to the result, for instance, skin colour and hair around. To tackle these issues, this paper proposed a SE-CNN model, which adopted squeeze-and-excitation attention and can be utilized to accurately classify the property of skin moles between benign and malignant without severe bias efficiently. SE-CNN obtained similar performance using fewer parameters than other state-of-the-art models we evaluated in the experiment, including Resnet, Dense Net, Efficient Net. In this paper, the attention mechanisms we included in the experiment were introduced. After that, we developed our own CNN model, adopted different attention modules into our model, and analysed the performance. Experiments of performance comparison with widely used neural networks models are shown after the development section. At last, we embedded our SE-CNN model into a skin cancer doctor web application for malignant & benign skin mole image classification and adopted an Albert NLP model for symptoms discussion with the user.

Citation Download Citation

Zian Song and Yi Zhou "Skin cancer classification based on CNN model with attention mechanism", Proc. SPIE 12179, Second International Conference on Medical Imaging and Additive Manufacturing (ICMIAM 2022), 1217917 (10 June 2022); https://doi.org/10.1117/12.2636772

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available

Members: $17.00

Non-members: $21.00 ADD TO CART

PROCEEDINGS
7 PAGES

DOWNLOAD PAPER SAVE TO MY LIBRARY

GET CITATION

RIGHTS & PERMISSIONS

Get copyright permission Get copyright permission on Copyright Marketplace

KEYWORDS

Skin cancer

Skin

Data modeling

Image classification

RGB color model

Convolution

Neural networks

RELATED CONTENT

Breast cancer diagnosis using SE-DNI model
Proceedings of SPIE (December 01 2021)

A CNN model used in a web page to classify...
Proceedings of SPIE (June 10 2022)

Traffic sign classification with dataset augmentation and convolutional neural network
Proceedings of SPIE (April 10 2018)

Deep multi-scale convolutional neural network for hyperspectral image classification
Proceedings of SPIE (April 10 2018)

Convolutional neural networks in skin cancer detection using spatial and...
Proceedings of SPIE (February 26 2019)

An ethnic costumes classification model with optimized learning rate
Proceedings of SPIE (August 14 2019)

Image classification based on self-attention convolutional neural network
Proceedings of SPIE (August 06 2021)

Subscribe to Digital Library

Receive Erratum Email Alert

Keywords/Phrases

Search In:

Publication Years