Classification of skin cancer based on composite scaling and attention

Kangli Xia; Qiang He

doi:10.1117/12.3005788

10 October 2023 Classification of skin cancer based on composite scaling and attention

Kangli Xia, Qiang He

Proceedings Volume 12799, Third International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2023); 1279953 (2023) https://doi.org/10.1117/12.3005788
Event: 3rd International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2023), 2023, Kuala Lumpur, Malaysia

Abstract

A suitable neural network can greatly improve the diagnostic performance of skin diseases. Based on the Compound Scaling Method, EfficientNet ^[8] combined with attention mechanism can achieve this goal. This article investigates a deep learning model that is more suitable for classifying skin cancer datasets, as well as the effectiveness of attention mechanisms in deep neural architecture. EfficientNet ^[8] based on the Compound Scaling Method is a more suitable model for skin cancer dataset classification, and the attention mechanism can enhance the attention to important features on the basic classification model. We compared the classification performance of Alex, VGG, ResNet, and EfficientNet ^[8] models on the skin cancer dataset HAM10000^[1]. Among the above four models, EfficientNet has better classification performance, achieving the highest accuracy of 91.65% on the HAM10000 dataset ^[1]. When EfficientNet ^[8] is combined with attention mechanism, performance is improved to a certain extent, achieving a precision of 92.98% on the HAM10000 dataset [1].

(2023) Published by SPIE. Downloading of the abstract is permitted for personal use only.

Citation Download Citation

Kangli Xia and Qiang He "Classification of skin cancer based on composite scaling and attention", Proc. SPIE 12799, Third International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2023), 1279953 (10 October 2023); https://doi.org/10.1117/12.3005788

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