Occlusal caries detection using random walker algorithm: A graph approach

• Published in: 2014 36th Annual International Conference of the IEEE Engineering in Medicine and Biology Society Vol. 2014; pp. 1929 - 1932
• Main Authors: Bampis, Christos G., Koutsouri, Georgia D., Berdouses, Elias, Tripoliti, Evanthia E., Iliopoulou, Dimitra, Koutsouris, Dimitrios, Oulis, Constantine, Fotiadis, Dimitrios I.
• Format: Conference Proceeding Journal Article
• Language: English
• Published: United States IEEE 01.01.2014
• Subjects: Accuracy; Algorithms; Classification algorithms; Color; Databases, Factual; Dental Caries - diagnosis; Dental Caries - pathology; Dentistry; Diagnosis, Computer-Assisted - methods; Gray-scale; Humans; Image color analysis; Image Processing, Computer-Assisted; Image segmentation; Models, Statistical; Photography, Dental; Reproducibility of Results; Software; Surface morphology; Transforms; Watersheds
• ISSN: 1094-687X; 1557-170X
• DOI: 10.1109/EMBC.2014.6943989

The aim of this work is to present a modification of the Random Walker algorithm for the segmentation of occlusal caries from photographic color images. The modification improves the detection and time execution performance of the classical Random Walker algorithm and also deals with the limitations and difficulties that the specific type of images impose to the algorithm. The proposed modification consists of eight steps: 1) definition of the seed points, 2) conversion of the image to gray scale, 3) application of watershed transformation, 4) computation of the centroid of each region, 5) construction of the graph, 6) application of the Random Walker algorithm, 7) smoothing and extraction of the perimeter of the regions of interest and 8) overlay of the results. The algorithm was evaluated using a set of 96 images where 339 areas of interest were manually segmented by an expert. The obtained segmentation accuracy is 93%.