Comparison of two deep-learning image reconstruction algorithms on cardiac CT images: A phantom study

Area of Science:

• Medical Imaging
• Radiology
• Image Processing

Background:

• Cardiac computed tomography (CT) protocols aim to balance image quality with radiation dose.
• Advanced image reconstruction algorithms are crucial for optimizing cardiac CT performance.
• Comparing novel algorithms like Precise IQ Engine (PIQE) and Advanced intelligent Clear-IQ Engine (AiCE) is essential for clinical advancement.

Purpose of the Study:

• To evaluate and compare the image quality performance of PIQE and AiCE algorithms.
• To assess algorithm performance across different radiation dose levels in cardiac CT.
• To determine the impact of PIQE and AiCE on noise reduction and lesion detectability.

Main Methods:

• Utilized the CT ACR 464 phantom with a prospective cardiac CT protocol at three dose levels (7.1/5.2/3.1 mGy).
• Reconstructed raw data using three levels (Mild, Standard, Strong) of both AiCE and PIQE algorithms.
• Analyzed image quality using noise power spectrum (NPS), task-based transfer function (TTF), and detectability index (d') for simulated cardiac lesions.

Main Results:

• PIQE consistently yielded lower noise magnitude values compared to AiCE across all dose levels.
• PIQE demonstrated a shift towards higher spatial frequencies in the NPS, indicating improved resolution.
• The detectability index (d') for simulated coronary lumen and non-calcified plaque was significantly higher with PIQE than AiCE.

Conclusions:

• PIQE outperformed AiCE in reducing noise, enhancing spatial resolution, and improving the detectability of simulated cardiac lesions.
• PIQE exhibits significant potential for enabling dose reduction strategies in cardiac CT acquisitions.
• The findings suggest PIQE as a valuable tool for optimizing cardiac CT image quality and patient safety.