Comparison of automatic and visual methods used for image segmentation in endodontics: A microCT study

Abstract

To calculate root canal volume and surface area in microCT images, an image segmentation by selecting threshold values is required, which can be determined by visual or automatic methods. Visual determination is influenced by the operator’s visual acuity, while the automatic method is done entirely by computer algorithms. Objective: To compare between visual and automatic segmentation, and to determine the influence of the operator’s visual acuity on the reproducibility of root canal volume and area measurements. Material and methods: Images from 31 extracted human anterior teeth were scanned with a μCT scanner. Three experienced examiners performed visual image segmentation, and threshold values were recorded. Automatic segmentation was done using the “Automatic Threshold Tool” available in the dedicated software provided by the scanner’s manufacturer. Volume and area measurements were performed using the threshold values determined both visually and automatically. Results: The paired Student’s t-test showed no significant difference between visual and automatic segmentation methods regarding root canal volume measurements (p=0.93) and root canal surface (p=0.79). Conclusion: Although visual and automatic segmentation methods can be used to determine the threshold and calculate root canal volume and surface, the automatic method may be the most suitable for ensuring the reproducibility of threshold determination.