Premium
SU‐F‐18C‐02: Evaluations of the Noise Power Spectrum of a CT Iterative Reconstruction Technique for Radiation Therapy
Author(s) -
Dolly S,
Chen H,
Anastasio M,
Mutic S,
Li H
Publication year - 2014
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.4889086
Subject(s) - imaging phantom , iterative reconstruction , region of interest , noise (video) , pixel , image noise , noise reduction , algorithm , mathematics , artificial intelligence , nuclear medicine , physics , computer science , optics , image (mathematics) , medicine
Purpose: To quantitatively assess the noise power spectrum (NPS) of the new, commercially released CT iterative reconstruction technique, iDose 4 from Philips, to compare it with filtered back‐projection techniques (FBP), and to provide clinical practice suggestions for radiation therapy. Methods: A uniform phantom was CT imaged with 120kVp tube potential over a range of mAs (250‐3333). The image sets were reconstructed using two reconstruction algorithms (FBP and iDose 4 with noise reduction levels 1, 3, and 6) and three reconstruction filters (standard B, smooth A, and sharp C), after which NPS variations were analyzed and compared on region of interest (ROI) sizes (16×16 to 128×128 pixels), ROI radii (0–65 mm), reconstruction algorithms, reconstruction filters, and mAs. Results: The NPS magnitude and shape depended considerably on ROI size and location for both reconstruction algorithms. Regional noise variance became more stationary as ROI size decreased, minimizing NPS artifacts. The optimal 32×32‐pixel ROI size balanced the trade‐off between stationary noise and adequate sampling. NPS artifacts were greatest at the center of reconstruction space and decreased with increasing ROI distance from the center. The optimal ROI position was located near the phantom's radial midpoint (∼40mm). For sharper filters, the NPS magnitude and the maximum magnitude frequency increased. Higher dose scans yielded lower NPS magnitudes for both reconstruction algorithms and all filters. Compared to FBP, the iDose 4 algorithm reduced the NPS magnitude while preferentially reducing noise at mid‐range spatial frequencies, altering noise texture. This reduction was more significant with increasing iDose 4 noise reduction level. Conclusion: Compared to pixel standard deviation, NPS has greater clinical potential for task‐based image quality assessment, describing both the magnitude and spatial frequency characteristics of image noise. While iDose 4 reduces the NPS magnitude more dramatically than FBP, further studies are needed to assess the impacts of noise texture alteration on its clinical usage.