Protocol Optimization Focused Scanner Characterization In CT: Noise and Spatial Resolution
Abstract
Purpose
Traditionally, CT scanner spatial resolution is characterized using non-clinically optimized scan modes (axial scans, with dose levels and kernels not suited for real patients). In this study, we demonstrate a clinically motivated approach for exploring how reconstruction field of view (RFOV) and kernels influence the spatial resolution and noise in CT imaging.
Methods
3 phantoms (Catphan 500 (The Phantom Laboratories), Triple Modality 3D Abdominal Phantom (CIRS), cadaveric head) were imaged in a whole-body CT scanner (GE Discovery CT750). All phantoms were scanned using the same tube voltage (120 kV), slice thickness (1.25mm), current (400mA), dose (35.3mGy), and scanning FOV (Large body). Each scan was reconstructed with 8 different kernels (soft tissue, standard, lung, chest, detail, bone, boneplus, edge) and 11 different reconstruction FOVs (diameters of 100, 150, 200, 250, 300, 320, 350, 360, 400, 450, 500 mm). Spatial resolution (line pairs/cm) and signal-to-noise ratio (SNR) were measured in a uniform region of interest (ROI) with a fixed physical area and location at the central slices of the Catphan. SNR was also measured in an ROI at the central slice of the cadaveric phantom and in three different ROIs (liver, high contrast, and low contrast) of the abdominal phantom.
Results
The SNR increases as the size of RFOV increases for all reconstruction kernels and phantoms due to increasing pixel size, which reduces noise through spatial averaging and noise correlation effect. However, the increase in SNR is compensated by the decrease in spatial resolution due to filtering, which suppresses fine details.
Conclusion
This work provides actionable information to aid radiologists in selecting reconstruction kernels for various clinical tasks that require differing levels of resolution and RFOV.