A quantitative validation of segmented colon in virtual colonoscopy using image moments

Area of Science:

• Medical Imaging
• Image Processing
• Computational Anatomy

Background:

• Accurate colon segmentation is a significant challenge in Computed Tomography Colonography (CTC).
• Existing methods require robust validation for clinical application.

Purpose of the Study:

• To accurately measure the segmented colon using advanced image processing techniques.
• To validate the accuracy and robustness of a novel colon segmentation method.

Main Methods:

• Retrospective analysis of 85 Computed Tomography Colonography (CTC) datasets.
• Application of three distance measurement techniques and volumetric overlap computation.
• Validation on 2D Multiplanar Reformation (MPR) images using Euclidean space measurements.

Main Results:

• Statistically demonstrated no significant difference between segmented and unsegmented colon measurements (p=0.6769).
• Achieved a high accuracy of 95.265±0.4551% via volumetric overlap computation.
• Quantitative validation on 2D MPR images confirmed the method's reliability.

Conclusions:

• The developed colon segmentation method is robust and accurate.
• The combination of validation techniques ensures reliable quantitative results.
• The approach is accepted for clinical use with a significance level of α=5%.