Impact of four kVp combinations available in a dual‐source CT on the spectral performance of abdominal imaging: A task‐based image quality assessment on phantom data

Abstract

To compare the spectral performance of four combinations of kVp available in a third generation dual-source CT (DSCT) on abdominal imaging. Methods: An image-quality phantom was scanned with a DSCT using four kVp pairs (tube "A" voltage/tube "B" voltage): 100/Sn150 kVp, 90/Sn150 kVp, 80/Sn150 kVp, and 70/Sn150 kVp, classic parameters and dose level for abdomen examination (CTDI vol : 11 mGy). The noise power spectrum (NPS) and the task-based transfer function (TTF) of two inserts were computed on virtual monochromatic images (VMIs) at 40/50/60/70 keV and for mixed, low-, and high-kVp images. Detectability index (d') was computed on VMIs and mixed images to model the detection task of liver metastasis (LM) and hepatocellular carcinoma (HCC). Iodine quantification accuracy was assessed using the Root Mean Square Deviation (RMSD iodine ) and the iodine bias (IB).Results: Noise magnitude decreased by −55%± 0% between 40 and 70 keV for all kVp pairs. Compared to 70/Sn150 kVp, noise magnitude was increased by 9% ± 0% with 80/Sn150 kVp, by 16% ± 1% with 90/Sn150 kVp and by 24%± 1% with 100/Sn150 kVp. The average NPS spatial frequency (f av ) shifted toward higher frequencies as energy level increased for all kVp pairs. Lowest f av values were found for 70/Sn150 kVp and highest for 100/Sn150 kVp. The value of TTF at 50% (f 50 ) shifted toward lower frequencies with increasing energy level. The highest f 50 values occurred for 100/Sn150 kVp and the lowest for 80/ Sn150 kVp. For both lesions, d' was highest for 70/Sn150 kVp and lowest for 100/Sn150 kVp. Compared to 70/Sn150 kVp, d' decreased by −6% ± 3% with 80/Sn150 kVp, by −11% ± 2% with 90/Sn150 kVp and by −13%± 2% with 100/ Sn150 kVp. For all acquisitions, the RSMD iodine and IB were the lowest for 100/ Sn150 kVp (0.29 ± 0.10 mg/ml and 0.88 ± 0.30 mg/ml, respectively) and increased when the tube "A" voltage decreased (2.34 ± 0.29 mg/ml for 70/Sn150 kVp and 7.42 ± 0.51 mg/ml respectively).

Phantom task‐based image quality assessment of three generations of rapid kV‐switching dual‐energy CT systems on virtual monoenergetic images

“…24 In this case, a conditioning of the ESF was performed using imQuest software to improve the TTF calculation and the TTF was often overestimated especially at low keV (40, 50, and 60 keV). 11,29,31 In addition, variations in TTFs may also be related to the composition of the inserts which move away from the clinical lesions and may therefore impact the Photoelectric/Compton discrimination for reconstructing VMI. To avoid these issues, in the present study we used higher contrast inserts containing iodine and calculated the TTF on 200 images from 10 acquisitions.…”

“…The low keV noise conditions and the low contrast of these inserts led to a relatively low CNR (CNR < 15) 24 . In this case, a conditioning of the ESF was performed using imQuest software to improve the TTF calculation and the TTF was often overestimated especially at low keV (40, 50, and 60 keV) 11,29,31 . In addition, variations in TTFs may also be related to the composition of the inserts which move away from the clinical lesions and may therefore impact the Photoelectric/Compton discrimination for reconstructing VMI.…”

Impact of an artificial intelligence deep‐learning reconstruction algorithm for CT on image quality and potential dose reduction: A phantom study

“…A task‐based image quality assessment was performed using the iQMetrix‐CT software developed by a working group from the French Society of Medical Physicists. No articles or reports have yet been published to describe the iQMetrix‐CT software; however, this software has been used in various studies 11,12,30 …”

“…No articles or reports have yet been published to describe the iQMetrix-CT software; however, this software has been used in various studies. 11,12,30…”

“…To use the same process than for the TTF calculation, the NPS was computed in 250 consecutive axial slices (50 slices for each of the 5 acquisitions) by placing 4 square regions of interest (ROIs) of 128 × 128 pixels (Figure 1a) using the same methodology as previously reported. 11,12,30 In the iQMetrix-CT software, the raw data NPS1D curves are fitted using an 11th-order polynomial.…”