Deakin University
Browse

File(s) under permanent embargo

A swarm optimization-based Kmedoids clustering technique for extracting melanoma cancer features

conference contribution
posted on 2017-01-01, 00:00 authored by Seyedamin Khatami, S Mirghasemi, Abbas KhosraviAbbas Khosravi, Chee Peng LimChee Peng Lim, Houshyar AsadiHoushyar Asadi, Saeid Nahavandi
© 2017, Springer International Publishing AG. Melanoma is a dangerous type of skin cancers. It is alarming to see the increase of this noxious disease in modern societies, however, it can be cured by surgical excision if it is detected early. In this paper, a swarm-based clustering technique for detecting melanoma is developed. Meaningful colour features from images are extracted, and a new objective function is introduced by applying an efficient and fast linear transformation to detect Melanoma. Specifically, the proposed technique consists of three main phases. The first phase is a pre-processing stage to organize data into proper attributes, while the subsequent two phases comprise iterative swarm optimisation procedures. The iterative swarm optimisation procedures involve a linear transformation to convert the existing colour components into a new colour space, formulation of the Kmedoids objective function, and error minimisation of the particle swarm optimisation (PSO) solutions. The Otsu threshold technique is utilised to provide binary images. The proposed technique is efficient and effective due to its linearity and simplicity.

History

Event

Neural Information Processing. International Conference (24th : 2017 : Guangzhou, China)

Volume

10637

Series

Lecture Notes in Computer Science

Pagination

307 - 316

Publisher

Springer

Location

Guangzhou, China

Place of publication

Cham, Switzerland

Start date

2017-11-14

End date

2017-11-18

ISSN

0302-9743

eISSN

1611-3349

ISBN-13

9783319700922

Language

eng

Publication classification

E1 Full written paper - refereed

Copyright notice

2017, Springer International Publishing AG

Title of proceedings

ICONIP 2017: Proceedings of the 24th International Conference on Neural Information Processing