Efficient Detection of Skin Cancer Using Deep Learning Techniques and a Comparative Analysis Study

Document Type

Book Chapter

Source of Publication

Lecture Notes in Electrical Engineering

Publication Date

6-3-2023

Abstract

Many skin lesions may result in the wrong diagnosis of skin cancer, leading to delays and ultimately making the cure impossible. Framed within this statement, this article proposes an efficient skin cancer detection model and compares the six pre-trained models, used for transfer learning in ISIC 2019 dataset. Three most common types of skin cancer—melanoma, nevus, and basal cell carcinoma—are classified by using the transfer learning on the pre-trained models of the ISIC 2019 dataset, to conclude the most accurate detection results with training and test accuracy of 99.73% and 93.79%, respectively.

ISBN

978-981-99-1251-3, 978-981-99-1252-0

ISSN

1876-1119

Publisher

Springer Nature Singapore

Volume

1028

First Page

203

Last Page

210

Disciplines

Computer Sciences

Keywords

Skin cancer detection, Deep convolutional neural network, Pre-trained models, Transfer learning

Indexed in Scopus

no

Open Access

no

Link to Full Text

Share

COinS