Objective: To evaluate the clinical application of a full model-based iterative reconstruction (MBIR) algorithm in the ultra-low-dose paranasal sinus CT imaging of children.
Materials and methods: In the first phase, 16 low-dose CT dacryocystography (DCG) (80 kV/64 mAs) scans were reconstructed with MBIR and filtered back-projection (FBP) to demonstrate noise reduction capability of MBIR. MBIR images were also compared with the images of 21 standard-dose paranasal sinus patients reconstructed with adaptive statistical iterative reconstruction (ASIR) algorithm. In the second phase, 14 pediatric tumors patients (images with ASIR in the initial scan) who came for follow-up paranasal sinus CT scan were prospectively enrolled with reduced radiation and MBIR algorithm. In both study phases, image noise and the contrast noise ratio (CNR) of sphenoid was measured; and subjective image quality was evaluated. CTDIvol and DLP were recorded, and effective dose calculated.
Results: The CTDIvol value for the DCG group was 63.9% lower than the standard-dose sinus group (1.09 ± 0.01 mGy vs. 3.02 ± 0.35 mGy). Compared with the ASIR reconstruction in the standard-dose sinus patient group, images with MBIR in the ultra-low-dose DCG group had 39.9% lower noise (9.5 ± 0.8HU vs. 15.8 ± 3.3HU) and 63.6% higher CNR (14.4 ± 4.7 vs. 8.8 ± 2.2), with similar subjective image quality score. For the tumor patients, 65.5% dose reduction was achieved. Subjective quality scores were similar between the initial and follow-up scans. Objective noise was significantly lower for the follow-up group.
Conclusion: MBIR provided equal or better image quality with significantly reduced radiation dose in paranasal sinus CT imaging of pediatric patients compared with standard-dose CT with ASIR algorithm.
Keywords: Children; Low-dose CT; Model-based iterative reconstruction algorithm (MBIR); Paranasal sinus; Radiation dose.