Scale-based scatter correction for computer-aided polyp detection in CT colonography.

Area of Science:

• Medical Imaging
• Radiology
• Computer-Aided Diagnosis

Background:

• CT colonography (CTC) is a minimally invasive colorectal cancer screening tool.
• Computer-aided detection (CAD) enhances polyp detection in CTC but is challenged by scatter effects from oral contrast agents.
• Pseudoenhancement from contrast agents artificially elevates CT attenuation, hindering CAD performance, particularly for submerged polyps.

Purpose of the Study:

• To develop and evaluate a scale-based correction method to minimize contrast agent scatter effects in CTC.
• To improve the sensitivity and reduce false positives of CAD for polyp detection in CTC.

Main Methods:

• A scale-based correction method was developed to subtract estimated scatter components from observed CT attenuations.
• The method adaptively identifies and corrects scatter effects based on local image structure and object scale.
• The correction was implemented as a preprocessing step in a CAD system and evaluated on 56 clinical CTC scans with 50 polyps (6-9 mm).

Main Results:

• Visual evaluation confirmed reduced CT attenuation of pseudoenhanced polyps to normal ranges without affecting air regions.
• CAD sensitivity increased by 24% for submerged polyps (at 10 false positives per scan) with the correction.
• Overall sensitivity for 6-9 mm polyps increased by 8% (at 10 false positives per scan) using the scatter-corrected CAD.

Conclusions:

• The scale-based scatter correction method effectively reduces pseudoenhancement artifacts in CTC.
• Implementing this correction as a preprocessing step significantly enhances CAD performance for polyp detection in CTC.
• The approach yields high sensitivity and a low false-positive rate for colorectal polyp detection using CTC.