Computational Detection of Pigmented Skin Cancer Images Using Machine Learning and Optimization Techniques

Skin cancer disease (SCD) is among the most prevalent kinds of cancer globally, with its incidence rising due to ultraviolet radiation, environmental alterations, and lifestyle factors. This article examines three primary types of skin cancer, causes, risk factors, prevention strategies, and treatment options: basal cell carcinoma, squamous cell carcinoma, and deadly melanoma. Firstly, we have tested ten machine learning (ML) algorithms on the featured dataset by integrating the clustering technique, i.e., principal component analysis (PCA). In the second stage, we applied ten ML algorithms, clustering techniques, and optimization techniques, i.e., grey wolf optimization (GWO). We observed improved predictive outcomes with the same dataset, achieving 94.05% accuracy, 92.82% precision, 99.75% sensitivity, 99.75% specificity, and 95.20% F1-score, which highlights the novelty of the hybrid model. Lastly, the ROC curves with the corresponding AUC values were included in the empirical analysis, yielding an enhanced AUC of 0.97. Through this analysis, the paper emphasizes the importance of early identification and active prevention of skin cancer-associated diseases and mortality. The obtained results indicate that the use of clustering techniques in conjunction with the optimization technique (i.e., GWO) and classification methods is quite advantageous.