Noise-Matched Blending Level Selection for 1024-Matrix CT Images Using Hybrid-Iterative Reconstruction: Comparison...

Area of Science:

• Medical Imaging
• Radiology
• Image Reconstruction

Background:

• Increasing computed tomography (CT) matrix size can increase image noise.
• Iterative reconstruction (IR) techniques aim to reduce noise and improve image quality.
• Adaptive statistical iterative reconstruction-Veo (ASiR-V) is a common IR technique.

Purpose of the Study:

• To determine the optimal noise-matched hybrid-IR blending level for 1024-matrix CT images.
• To achieve noise levels equivalent to 512-matrix images while preserving image quality.
• To evaluate the impact of noise-matched hybrid-IR on image sharpness and overall quality.

Main Methods:

• A phantom study identified the noise-matched hybrid-IR blending level (60%) for 1024-matrix images.
• A retrospective study included 63 patients undergoing pancreatic protocol CT.
• Images were reconstructed at 512-matrix (ASiR-V 30%), 1024-matrix (ASiR-V 30%), and 1024-matrix (noise-matched ASiR-V 60%).

Main Results:

• The noise-matched 1024-matrix images (60% ASiR-V) exhibited significantly lower background noise than 1024-matrix images with 30% ASiR-V.
• Edge-rise slope (ERS) was significantly higher in 1024-matrix images (both 30% and 60% ASiR-V) compared to 512-matrix images.
• Overall image quality was significantly superior for the noise-matched 1024-matrix images compared to the other groups.

Conclusions:

• Applying a noise-matched hybrid-IR blending level (60% ASiR-V) allows for 1024-matrix CT images to maintain image quality.
• This technique improves edge sharpness without increasing noise compared to lower-matrix images.
• Noise-matched hybrid-IR is a promising method for enhancing image quality in high-resolution CT.