Melanoma classification by 3D colour-texture feature and neural network with improved computational complexity using PCA
by Mohd Firoz Warsi; Ruqaiya Khanam; Usha Chauhan; Suraj Kamya
International Journal of Medical Engineering and Informatics (IJMEI), Vol. 14, No. 4, 2022

Abstract: The most severe kind of skin cancer is malignant melanoma. It can grow anywhere on the body. Its exact cause is still unclear but typically it is caused by ultraviolet exposure from sun or tanning beds. Its detection plays a very significant role because if detected early then it is curable, before the spread has begun. In this paper, a computationally improved [using principal component analysis (PCA)] feature extraction method named 3D colour texture feature (CTF) is represented which is well discriminative. For classification of melanoma from dermoscopic images, a comparison of different types of machine-learning classification algorithms is evaluated, out of which back propagation neural network (NN) classifier outperforms all other and produce best results, i.e., accuracy = 98.5%, sensitivity = 99.4%, specificity = 95.0%. Obtained results are even better than benchmarking results of PH² dataset. Comparisons of results with other similar novel works are also discussed.

Online publication date: Tue, 05-Jul-2022

The full text of this article is only available to individual subscribers or to users at subscribing institutions.

 
Existing subscribers:
Go to Inderscience Online Journals to access the Full Text of this article.

Pay per view:
If you are not a subscriber and you just want to read the full contents of this article, buy online access here.

Complimentary Subscribers, Editors or Members of the Editorial Board of the International Journal of Medical Engineering and Informatics (IJMEI):
Login with your Inderscience username and password:

Username:        Password:          Forgotten your password?

Want to subscribe?
A subscription gives you complete access to all articles in the current issue, as well as to all articles in the previous three years (where applicable). See our Orders page to subscribe.

If you still need assistance, please email subs@inderscience.com