Three-dimensional CT imaging in extensor tendons using deep learning reconstruction: optimal reconstruction...

Area of Science:

• Radiology
• Medical Imaging
• Computational Imaging

Background:

• Three-dimensional computed tomography (3D CT) is crucial for visualizing extensor tendons.
• Optimizing reconstruction parameters is essential for improving image quality and diagnostic accuracy.
• Deep learning reconstruction (DLR) offers potential advancements over traditional iterative reconstruction methods.

Purpose of the Study:

• To evaluate optimal reconstruction parameters for extensor tendon 3D CT using DLR.
• To assess the influence of tube current on image quality with DLR.
• To compare DLR (AiCE) with adaptive iterative dose reduction three dimensional (AIDR 3D) for tendon imaging.

Main Methods:

• Phantom study using a 3 mm rod simulating extensor tendons at varying tube currents (50-250 mA).
• Clinical study involving CT scans of hand tendons from nine hands (eight patients).
• Image reconstruction using AiCE (body, body sharp, brain CTA, brain LCD) and AIDR 3D; image quality assessed via contrast-to-noise ratio (CNR) and coefficient of variation (CV).

Main Results:

• In phantom studies, AiCE parameters demonstrated comparable or superior CNR_LO to AIDR 3D at 200 mA.
• Clinical studies showed AiCE brain CTA achieved higher CNR and lower CV compared to other AiCE parameters and AIDR 3D.
• AiCE without dose reduction potentially improves extensor tendon 3D CT image quality.

Conclusions:

• AiCE brain CTA is more suitable for extensor tendon 3D CT than other AiCE parameters.
• DLR, particularly AiCE brain CTA, offers improved image quality for extensor tendon visualization.
• Further research into AiCE without dose reduction may enhance tendon imaging efficacy.