Psoriasis Lesion Detection Using Hybrid Seeker Optimization-based Image Clustering

Background: In recent years, there has been a massive increase in the number of people suffering from psoriasis. For proper psoriasis diagnosis, psoriasis lesion segmentation is a prerequisite for quantifying the severity of this disease. However, segmentation of psoriatic lesions cannot be evaluated just by visual inspection as they exhibit inter and intra variability among the severity classes. Most of the approaches currently pursued by dermatologists are subjective in nature. The existing conventional clustering algorithm for objective segmentation of psoriasis lesion suffers from limitations of premature local convergence. Objective: An alternative method for psoriatic lesion segmentation with objective analysis is sought in the present work. The present work aims at obtaining optimal lesion segmentation by adopting an evolutionary optimization technique that possesses a higher probability of global convergence for psoriasis lesion segmentation. Methods: A hybrid evolutionary optimization technique based on the combination of two swarm intelligence algorithms, namely Artificial Bee Colony and Seeker Optimization algorithm, has been proposed. The initial population for the hybrid technique is obtained from the two conventional local- based approaches, i.e., Fuzzy C-means and K-means clustering algorithms. Results: The initial population selection from the convergence of classical techniques reduces the effect of population dynamics on the final solution and hence yields precise lesion segmentation with a Jaccard Index of 0.91 from 720 psoriasis images. Conclusion: The performance comparison reflects the superior performance of the proposed algorithm over other swarm intelligence and conventional clustering algorithms.