Intelligent Skin Cancer Detection Using Enhanced Particle Swarm Optimization

Abstract

This research undertakes intelligent skin cancer diagnosis based on dermoscopy images using several variants of the Particle Swarm Optimization (PSO) algorithm for feature optimization. Since the identification of the most significant discriminative characteristics of the benign and malignant skin lesions plays an important role in robust skin cancer detection, the proposed PSO algorithms are employed for feature optimization. Specifically, the overall system contains multiple steps, i.e. pre-processing (noise removal), segmentation, feature extraction from both skin and lesion regions, proposed PSO based feature selection and classification. After extracting a large number of raw shapes, colour and texture features from the lesion areas, feature selection is conducted to identify the most discriminating significant feature subsets. Besides PSO and Genetic Algorithm (GA) based feature optimization, a total of four novel PSO variant algorithms, i.e. hybrid learning PSO (HLPSO), a PSO variant model (PSOVA), adaptive coefficient PSO (ACPSO), and random coefficient PSO (RCPSO), have been proposed for feature selection. Diverse search strategies are proposed in these models to mitigate premature convergence problems of the original PSO algorithm. Single and ensemble classifiers have been employed to perform benign and malignant lesion classification. Evaluated with multiple skin lesion and UCI databases and diverse unimodal and multimodal benchmark functions, the proposed PSO variants show superior performances over those of other advanced and classical search methods for identifying discriminative features that facilitate benign and malignant lesion classification as well as for solving diverse optimization problems with different landscapes. The Wilcoxon rank sum test is adopted to further ascertain superiority of the proposed algorithms over other methods statistically.

Date of Award	1 Jun 2019
Original language	English
Awarding Institution	Northumbria University