CT iterative reconstruction algorithms: a task-based image quality assessment.

Area of Science:

• Medical Imaging
• Radiology
• Computational Imaging

Background:

• Filtered Back Projection (FBP) is a standard CT reconstruction method.
• Iterative Reconstruction (IR) algorithms offer potential for improved image quality and reduced radiation dose.
• Evaluating the performance of different IR generations across vendors is crucial for clinical optimization.

Purpose of the Study:

• To compare the dose performance and image quality of FBP and two generations of IR algorithms from major CT vendors.
• To assess the effectiveness of IR in detecting simulated liver lesions and calcifications at various dose levels.

Main Methods:

• Four CT systems with Hybrid/Statistical IR (H/SIR) and Model-Based IR (MBIR) were used.
• ACR phantom scans were performed at five dose levels.
• Image quality was assessed using Noise Power Spectrum (NPS), Task-Based Transfer Function (TTF), and a detectability index (d') for simulated lesions.

Main Results:

• Siemens H/SIR showed the highest detectability index (d') for both large and small simulated lesions.
• MBIR allowed for significant dose reductions compared to H/SIR across vendors, particularly for Philips and GE.
• Canon's MBIR showed dose reduction for small features but a dose increase for large features compared to H/SIR.

Conclusions:

• Multivendor IR algorithm comparison reveals vendor-specific evolutionary improvements.
• The detectability index (d') is a robust metric for optimizing CT image quality and dose.
• Task-based image quality assessment provides clinically relevant insights beyond complex metrics.