A fully automatic CAD-CTC system based on curvature analysis for standard and low-dose CT data

Area of Science:

• Medical Imaging
• Radiology
• Computer-Aided Diagnosis

Background:

• Computed tomography colonography (CTC) is a noninvasive screening method for colorectal polyps.
• Increasing data volumes in CTC challenge manual analysis, leading to time consumption and variability.
• Development of automated tools is crucial for efficient and reliable polyp detection.

Purpose of the Study:

• To implement and detail a fully integrated computer-aided detection system for computed tomography colonography (CAD-CTC).
• To enable robust identification of clinically significant colorectal polyps within CT data.
• To address the challenges of manual analysis in CTC due to increasing data complexity.

Main Methods:

• The CAD-CTC system employs a multistage approach.
• Key components include automatic colon segmentation, candidate surface extraction, feature extraction, and classification.
• The system was evaluated on both synthetic and real patient CT datasets.

Main Results:

• The CAD-CTC system achieved 100% sensitivity for polyps >10 mm.
• Sensitivity was 92% for polyps 5-10 mm and 57.14% for polyps <5 mm.
• An average of 3.38 false positives per dataset was recorded.

Conclusions:

• The developed CAD-CTC system demonstrates robust performance in detecting colorectal polyps.
• The system shows high accuracy for clinically significant polyps, aiding CTC interpretation.
• Further evaluation on diverse datasets could enhance its clinical utility.