Color diffusion model for active contours - an application to skin lesion segmentation.

Area of Science:

• Computer Vision
• Medical Image Analysis
• Color Image Processing

Background:

• Existing diffusion models are primarily for grayscale images, limiting their application in color medical imaging.
• Active contours require robust external energy functions for accurate segmentation, especially in complex color images.

Purpose of the Study:

• To propose and validate a novel diffusion model for color images.
• To utilize this model as an external energy function for active contours in image segmentation.
• To specifically address the segmentation of skin lesions in medical color images.

Main Methods:

• Developed a diffusion model based on the first-order moment of the correlation integral using CIE Lab color space ΔE distances.
• Implemented a multi-scale approach for active contours, computing diffusion independently at various scales.
• Validated the model on synthetic color textures and medical images of melanoma.

Main Results:

• The proposed diffusion model effectively processes color image information.
• Multi-scale computation enhances the robustness of the active contour segmentation.
• Successful validation on both synthetic and real-world medical data.

Conclusions:

• The novel diffusion model is suitable for color image segmentation using active contours.
• This approach shows significant potential for accurate skin lesion segmentation in clinical settings.
• The method advances the application of diffusion models in medical image analysis.