Psoriasis Lesion Detection Using Hybrid Seeker Optimization-based Image Clustering.

Area of Science:

• Dermatology
• Medical Imaging
• Computational Intelligence

Background:

• Psoriasis prevalence is increasing, necessitating accurate methods for disease severity quantification.
• Psoriatic lesion segmentation is crucial for objective diagnosis, but visual inspection is subjective due to lesion variability.
• Conventional clustering algorithms for segmentation suffer from local convergence issues, limiting their effectiveness.

Purpose of the Study:

• To develop an objective method for psoriatic lesion segmentation using evolutionary optimization.
• To achieve optimal lesion segmentation with a higher probability of global convergence.

Main Methods:

• A hybrid evolutionary optimization technique combining Artificial Bee Colony and Seeker Optimization algorithms was proposed.
• Initial populations were derived from Fuzzy C-means and K-means clustering algorithms.
• This approach aimed to mitigate the impact of population dynamics on segmentation accuracy.

Main Results:

• The proposed algorithm achieved precise lesion segmentation with a Jaccard Index of 0.91 on 720 psoriasis images.
• Initial population selection from classical techniques improved the final segmentation outcome.
• The method demonstrated superior performance compared to other swarm intelligence and clustering algorithms.

Conclusions:

• The developed hybrid evolutionary algorithm offers a more objective and accurate approach to psoriatic lesion segmentation.
• This technique addresses the limitations of conventional methods, improving diagnostic capabilities for psoriasis.
• The findings suggest a significant advancement in computational dermatology for disease assessment.