A straightforward approach to computer-aided polyp detection using a polyp-specific volumetric feature in CT...

Area of Science:

• Medical Imaging
• Computer-Aided Diagnosis
• Gastroenterology

Background:

• Colorectal cancer screening relies on accurate polyp detection.
• CT colonography (CTC) is an alternative to optical colonoscopy (OC).
• Computer-aided detection (CAD) can enhance polyp identification in CTC.

Purpose of the Study:

• To develop and evaluate a straightforward computer-aided polyp detection (CAD) scheme for CTC.
• To assess the performance of the CAD scheme for clinically relevant polyps (≥ 6 mm).

Main Methods:

• A CAD scheme was developed involving colon wall segmentation and a polyp-specific volumetric filter.
• The system counted and thresholded cluster volume sizes to identify polyps.
• Data from 65 patients (103 polyps) underwent training and testing, with OC polyp sizes as reference.

Main Results:

• The CAD scheme achieved 100% per-polyp sensitivity with 8.5 false positives/patient on the training dataset.
• On the test dataset, sensitivity was 93.3% with 7.7 false positives/patient for polyps ≥ 6 mm.
• The system demonstrated high detection performance with easily optimizable parameters.

Conclusions:

• The developed CAD scheme offers a straightforward and effective approach for polyp detection in CTC.
• The method shows high sensitivity and a manageable false positive rate for clinically significant polyps.
• This CAD scheme has potential for improving colorectal cancer screening through enhanced polyp detection.