Improved computer-aided detection of small polyps in CT colonography using interpolation for curvature estimation

Area of Science:

• Medical imaging analysis
• Computational geometry
• Diagnostic imaging

Background:

• Surface curvatures are crucial for polyp detection in CT colonography (CTC).
• Standard curvature computation methods struggle with small polyps and those near haustral folds, reducing detection accuracy.
• Accurate curvature estimation is vital for effective computer-aided detection (CAD) systems.

Purpose of the Study:

• To analyze the impact of image interpolation on curvature estimation accuracy for thin structures in CTC.
• To apply interpolation techniques to enhance the computer-aided detection of small polyps in CTC.
• To address limitations of traditional curvature computation in polyp analysis.

Main Methods:

• Investigated the effect of image interpolation on curvature estimation for thin structures.
• Evaluated linear, quadratic B-spline, and cubic B-spline interpolation methods.
• Tested the application on simulated and clinical CTC data, focusing on small polyp detection.

Main Results:

• Image interpolation yielded more accurate curvature values for simulated data.
• Interpolation facilitated the isolation of polyps located near haustral folds in clinical data.
• Tested interpolations significantly improved the sensitivity of small polyp detection in a large clinical dataset.

Conclusions:

• Image interpolation enhances the accuracy of curvature estimation for thin structures.
• This improvement directly translates to better computer-aided detection of small polyps in CT colonography.
• Interpolation offers a valuable technique for refining CAD systems in medical imaging.