Application of Deep Learning-Based Denoising Technique for Radiation Dose Reduction in Dynamic Abdominal CT:...

Area of Science:

• Radiology
• Medical Imaging
• Artificial Intelligence in Medicine

Background:

• Computed tomography (CT) is crucial for abdominal imaging.
• Reducing radiation dose in CT is a significant clinical goal.
• Iterative reconstruction (IR) techniques have improved CT image quality but dose reduction remains challenging.

Purpose of the Study:

• To evaluate the efficacy of a deep learning-based denoising (DLD) algorithm in maintaining abdominal CT image quality at a reduced radiation dose.
• To compare DLD-processed, reduced-dose CT images with standard-dose CT images reconstructed using conventional hybrid iterative reconstruction (IR).

Main Methods:

• A study involving 100 patients: 50 with standard-dose CT (hybrid IR) and 50 with reduced-dose CT (hybrid IR + DLD at low, medium, high strengths).
• Image noise was quantified by measuring the standard deviation of attenuation in the liver parenchyma.
• Contrast-to-noise ratio (CNR) of the portal vein was calculated, and lesion conspicuity of abdominal masses was assessed using a 5-point scale.

Main Results:

• Reduced-dose CT with high-strength DLD (DLD-H) demonstrated significantly lower image noise and higher CNR compared to standard-dose CT with hybrid IR (P < 0.0001 for noise, P = 0.0019 for CNR).
• Medium-strength DLD (DLD-M) on reduced-dose CT showed comparable image noise and CNR to standard-dose CT with hybrid IR (P > 0.99).
• Lesion conspicuity was significantly improved with both DLD-H and DLD-M compared to hybrid IR on standard-dose CT (P < 0.05).

Conclusions:

• Deep learning-based denoising (DLD) enables significant radiation dose reduction in abdominal CT while preserving or enhancing image quality.
• The DLD algorithm, particularly at medium and high strengths, provides a viable method for improving image quality in reduced-dose abdominal CT.
• DLD represents a promising advancement for safer and effective abdominal CT imaging.