Technical Note: Measuring contrast- and noise-dependent spatial resolution of an iterative reconstruction method in...

Area of Science:

• Medical Imaging
• Radiology
• Image Reconstruction

Background:

• Iterative reconstruction (IR) in computed tomography (CT) offers noise reduction but its spatial resolution is contrast- and noise-dependent due to nonlinear regularization.
• Precise spatial resolution measurements in CT are difficult at very low contrast levels because of severe noise contamination.

Purpose of the Study:

• To measure the spatial resolution of a commercial iterative reconstruction (IR) method in CT.
• To evaluate the impact of low contrast and dose levels on IR spatial resolution.
• To assess the utility of ensemble-averaged images from repeated scans for spatial resolution assessment.

Main Methods:

• A low-contrast phantom with varying HU levels was scanned using a 128-slice CT scanner at multiple dose levels (16, 8, and 4 mGy).
• Images were reconstructed using filtered-backprojection (FBP) and a commercial IR method (SAFIRE).
• Repeated scans (100x) were ensemble-averaged to calculate the modulation transfer function (MTF) for spatial resolution assessment.

Main Results:

• Filtered-backprojection (FBP) spatial resolution remained constant across contrast and dose levels.
• The iterative reconstruction (IR) method showed degraded spatial resolution compared to FBP as contrast or dose decreased.
• At 16 mGy, MTF50% for IR decreased from 98.4% (21 HU) to 82.1% (7 HU) as contrast lowered.
• At 21 HU, MTF50% for IR decreased from 98.4% (16 mGy) to 85.6% (4 mGy) as dose lowered.

Conclusions:

• Ensemble averaging of repeated CT scans provides a viable method for measuring IR spatial resolution, even at very low contrast levels.
• The evaluated commercial IR method demonstrated a reduction in spatial resolution under low-contrast and high-noise conditions.