The impact of a novel deep learning reconstruction algorithm on image quality in ultralow-dose CT: a quantitative...

Area of Science:

• Medical Imaging
• Artificial Intelligence in Radiology
• Computed Tomography

Background:

• Ultralow-dose computed tomography (CT) imaging offers reduced radiation exposure but often suffers from increased image noise and decreased contrast-to-noise ratio (CNR).
• Deep learning image reconstruction (DLIR) algorithms show promise for improving image quality in low-dose CT.
• Evaluating the performance of DLIR for noise reduction, CNR enhancement, and low iodine concentration detection is crucial for clinical application.

Purpose of the Study:

• To assess the efficacy of a novel DLIR algorithm in noise reduction for ultralow-dose CT.
• To evaluate the impact of DLIR on contrast-to-noise ratio (CNR) at various iodine concentrations.
• To determine the capability of DLIR in detecting low iodine concentrations under ultralow-dose CT conditions.

Main Methods:

• A phantom with varying iodine concentrations was scanned using ultralow-dose CT at different tube voltages.
• Images were reconstructed using filtered back projection (FBP), iterative reconstruction (IR), and the novel DLIR algorithm.
• Objective metrics (noise, CNR, CT value accuracy) and subjective image quality were evaluated.

Main Results:

• DLIR significantly reduced noise and increased CNR compared to FBP and IR across all tested tube voltages (p < 0.001).
• The DLIR algorithm successfully enhanced CNR at low iodine concentrations (1.25 mg/mL) and enabled continuous detection (CNR ≥ 3) at all voltages.
• DLIR preserved diagnostic image quality and low-contrast detectability under ultralow-dose conditions, with up to 84.7% noise reduction compared to FBP.

Conclusions:

• The DLIR algorithm effectively enhances image quality and improves low iodine concentration detection in ultralow-dose CT phantom studies.
• DLIR maintains diagnostic image quality and CNR even with significant radiation dose reduction (up to 94%).
• These findings support further clinical research for optimizing dosage and potentially reducing iodine contrast agent use.