Occlusal caries detection using random walker algorithm: a graph approach.

Area of Science:

• Medical Imaging
• Computer Vision
• Dental Diagnostics

Background:

• Dental caries detection from images is crucial for diagnosis.
• Traditional segmentation algorithms face challenges with image quality and specific features of occlusal caries.
• The Random Walker algorithm offers a promising approach but requires optimization for this application.

Purpose of the Study:

• To present a modified Random Walker algorithm for segmenting occlusal caries.
• To improve detection accuracy and execution time compared to the classical algorithm.
• To address limitations specific to photographic color images of dental caries.

Main Methods:

• A novel eight-step process including seed point definition, grayscale conversion, watershed transformation, centroid computation, graph construction, Random Walker application, region smoothing, and result overlay.
• Evaluation involved segmenting 339 regions of interest across 96 images by an expert.

Main Results:

• The modified Random Walker algorithm achieved a segmentation accuracy of 93%.
• The proposed modifications enhanced detection performance and reduced execution time.
• The algorithm effectively handled the complexities of dental photographic images.

Conclusions:

• The modified Random Walker algorithm is effective for accurate occlusal caries segmentation.
• This method offers a significant improvement over the classical algorithm for dental image analysis.
• The approach shows potential for enhancing computer-aided diagnosis in dentistry.