MENU

Click here to

×

Are you sure ?

Yes, do it No, cancel

Matching Convolution Kernels and Iterative Reconstruction for Quantitative Accuracy and Noise Power Spectrum in Dual-Layer and Dual-Source Spectral CT

G Anthony*, Y Liang, Indiana University, Indianapolis, IN

Presentations

(Sunday, 7/12/2020)   [Eastern Time (GMT-4)]

Room: AAPM ePoster Library

Purpose: To identify comparable reconstruction techniques in two advanced spectral CT scanners by examining noise power spectra (NPS) and material decomposition.

Methods: The CT American College of Radiology phantom was scanned using standard clinical abdominal protocols on a dual-layer (DL) (Philips iQon, Best, The Netherlands) and dual-source (DS) (Siemens SOMATOM Force, Erlangen, Germany) scanner. Several body (Br) reconstruction kernels and a quantitative (Qr) kernel for the DS scanner were compared with the default body (B) kernel for the DL scanner. Multiple levels of iterative reconstruction were tested in both scanners. The NPS, effective atomic number (Z(eff)), and density (?) were derived for various scan and reconstruction techniques.

Results: The DS scanner Br44 kernel NPS most closely matched that of the DL scanner B kernel. Iterative reconstruction induced similar noise reduction and center frequency shifts, quantified by noise magnitude ratio (NMR) and central frequency ratio (CFR) of the NPS curves. For the DL system, NMR = 0.75 and 0.58, CFR = 0.91 and 0.82 for iDose 2 and iDose 4, respectively; for the DS system, NMR = 0.84 and 0.58, CFR = 0.92 and 0.83 for ADMIRE 1 and ADMIRE 3, respectively. When using the default iterative reconstruction settings (iDose 4, ADMIRE 3), the DS system reconstructions exhibited a slightly sharper noise texture (center frequency = 0.025 mm-1 greater) and reduced noise magnitude (NMR = 0.90–0.93) over the DL system. Z(eff) and ? measurements for both scanners were accurate within 6%. Iterative reconstruction techniques and the Qr kernel did not alter measurements of Z(eff) or ?.

Conclusion: When using matched reconstruction techniques, DL and DS scanners achieve comparable levels of noise magnitude and texture at the default iterative level. Quantitation of Z(eff) and ? in tissue-equivalent material is comparable in DL and DS scanners and is unaffected by iterative reconstruction.

Keywords

Dual-energy Imaging, Noise Power Spectrum, Quantitative Imaging

Taxonomy

IM- CT: Dual Energy and Spectral

Contact Email