Premium
WE‐C‐110‐08: A Novel Phantom for CT Performance Assessment: Towards a Task‐Based Measure of Image Quality
Author(s) -
Richard S,
Samei E
Publication year - 2011
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 - 10.1118/1.3613347
Subject(s) - imaging phantom , image quality , optical transfer function , iterative reconstruction , medical imaging , image resolution , noise (video) , projection (relational algebra) , optics , signal to noise ratio (imaging) , computer science , slab , artificial intelligence , materials science , physics , image (mathematics) , algorithm , geophysics
Purpose: The advent of CT iterative reconstructions in the clinic has posed an increasing challenge on the community in our ability to assess image quality due to increased non‐stationarity and non‐linearity of reconstructed images. The purpose of this study is to investigate the potential of a novel phantom aimed to enable a robust assessment of image quality of CT images including iterative reconstructions. Method and Materials: A cylindrical phantom was built from a series of slabs with removable rings to simulate diffrerent patient sizes. The “noise slab” consists of a uniform piece of acrylic for the noise‐power spectrum (NPS) measurement. The “resolution slab” consists of an array of spheres of various materials to simulate a range of contrast levels found in CT images. Each set of spheres can be used to measure the modulation transfer function (MTF) as a function contrast, dose, and location. The “dose slab” enables the incorporation of ion chambers, while the last slab contains cylindrical inserts for CNR measurements. The phantom was scanned on a 64 slice CT and reconstructed with filtered‐back projection (FBP) and a model‐based iterative reconstruction algorithm (MBIR). Images were used to derive the relevant image quality and dose metrics. Results: The NPS measurement enabled the characterization of different textures between FBP and MBIR. Contrast and dose were found to have a significant impact on the MTF for the MBIR data. The task‐based detectability index computed from the MTF and NPS agreed qualitatively with image quality and yielded quantitative estimation of dose reduction provided with iterative reconstruction algorithms. Conclusions: Initial results showed that a phantom can be used to measure imaging performance of non‐linear reconstruction algorithms. Results suggested that a task‐based approach to system performance assessment may be essential to properly compare image quality between different protocols when employing iterative reconstruction algorithms. The authors have received funding support from GE healthcare.