Novel Method for Border Irregularity Assessment in Dermoscopic Color Images

Area of Science:

• Dermatology
• Medical Imaging
• Computer Vision

Background:

• Border irregularity is a key indicator for predicting skin lesion malignancy.
• Accurate assessment of border characteristics is crucial for differentiating benign from malignant skin lesions.

Purpose of the Study:

• To develop an efficient automatic algorithm for detecting border irregularities in dermoscopic images.
• To enhance the diagnostic accuracy of the ABCD rule of dermoscopy through computational analysis.

Main Methods:

• The study involved image enhancement, lesion segmentation, borderline calculation, and irregularity detection.
• A novel method utilizing lesion rotation and borderline division was implemented to determine exact lesion boundaries.
• The algorithm was tested on 350 dermoscopic images.

Main Results:

• The developed algorithm achieved a diagnostic accuracy of 92% in detecting border irregularities.
• The computational approach effectively captured lesion irregularities, providing reliable data for skin mole examination.
• Improved classification results were observed compared to existing state-of-the-art methods.

Conclusions:

• Computer-aided systems offer a practical tool for dermoscopic image assessment.
• The proposed algorithm is suitable for both research and clinical applications in dermatology.
• The methodology has potential applications in other medical image analysis fields, such as CT and MRI.