Dependence of Resolution Properties on Display Field-of-view Size in Deep Learning-based Reconstruction Computed...

Area of Science:

• Medical Imaging
• Radiology
• Image Reconstruction

Background:

• Deep learning-based reconstruction (DLR) offers enhanced image quality in computed tomography (CT) over filtered back projection (FBP).
• Assessing small vascular structures in head CT angiography requires high spatial resolution.
• The impact of display field-of-view (D-FOV) on DLR spatial resolution is not fully understood.

Purpose of the Study:

• To investigate how display field-of-view (D-FOV) size affects the spatial resolution of deep learning-based reconstruction (DLR) images.
• To compare the spatial resolution properties of DLR, iterative reconstruction (IR), and filtered back projection (FBP) under varying D-FOV conditions.
• To evaluate the impact on detecting small vascular signals in CT angiography.

Main Methods:

• Utilized a vascular phantom with 0.5 mm and 1.0 mm diameter circular signals filled with contrast medium.
• Reconstructed images using FBP, IR, and DLR with standard denoising.
• Evaluated three D-FOV sizes (100 mm, 200 mm, 400 mm) and assessed spatial resolution via task transfer function (TTF) measurements.

Main Results:

• Filtered back projection (FBP) showed consistent TTF values across all D-FOV sizes.
• Iterative reconstruction (IR) and DLR exhibited lower TTF for 0.5 mm signals compared to 1.0 mm signals.
• TTF values in IR and DLR reconstructions decreased as D-FOV size increased.

Conclusions:

• Spatial resolution, measured by TTF, is negatively impacted by smaller signal sizes in IR and DLR.
• Increasing D-FOV size progressively degrades spatial resolution in IR and DLR CT images.
• These findings are critical for optimizing DLR and IR parameters in head CT angiography for small vessel visualization.