Ethics
The study was approved by the Swedish Ethical Review Authority (Dnr 2020-06114.).
EID-CT and PCD-CT image acquisition and reconstructions
Five cadaveric hearts were positioned in a chest phantom (N1 Lungman; Kyoto Kagaku Co. Ltd, Japan) and scanned in both an EID-CT (SOMATOM Force; Siemens Healthineers, Forchheim, Germany) as well as in a PCD-CT prototype (SOMATOM Count Plus; Siemens Healthineers, Forchheim Germany).
ECG-gating was not available in the PCD-CT. The vendor-provided spiral cardiac CAC Score protocol on the EID-CT is ECG-gated and ECG dose modulated. Thus, the expressed CTDIvol is based on the average tube current of the whole scan including both low and full dose cardiac phases. As only the full dose phases were used for image reconstructions, the expressed CTDIvol on the EID-CT, would render a too low radiation output on the PCD-CT. In order to get non-gated spiral protocols with an equal radiation output on both systems, we therefore performed the following procedure on the chest phantom before the first examination:
1. A CAC Score ECG-gated spiral scan of the phantom was made using a synthetic ECG on the EID-CT. Automatic exposure control (CARE Dose4D, Siemens Healthineers), vendor recommended Q. ref. mAs of 80 and ECG dose modulation was used. Image reconstructions were made during the full dose phase at 70% of the cardiac cycle using a 160 mm FoV. The dedicated Calcium Score kernel Sa36, as well as a 3 mm slice thickness with 1.5 mm increment were applied, as recommended by the vendor.
2. In total, nine non-ECG-gated spiral test scans were made on the EID-CT with automatic mAs exposure control (CARE Dose4D, Siemens Healthineers) using different Q. ref. mAs settings between 10 and 50. All scans were reconstructed in the same manner as the ECG-gated spiral.
3. The noise level in each test scan was determined by the placement of equal sized regions of interest (ROI) in the slices with the same slice position and at the same location in the image. By comparing the standard deviation (SD) in the non-ECG-gated scans with the SD in the ECG-gated scan a suitable Q. ref. mAs setting was found, i.e. the one rendering equal image noise (35 mAs). This Q. ref. mAs setting was then applied in the non-ECG-triggered thorax protocol used for all the following cadaveric heart scans at the EID-CT within the study.
Scans of the cadaveric hearts at the PCD-CT were made directly after the scans on the EID-CT. By matching the CTDIvol between the scans as closely as possible (CTDIvol varying between 0.85 and 1.14 mGy between the different cadaveric hearts) a similar radiation output was ensured. All scans within the study were performed with a spiral protocol, using a tube potential of 120 kV. In order to evaluate the reproducibility on both systems, the phantom was scanned once and then manually repositioned, after which it was scanned again.
Reconstructions were performed using quantitative kernels for calcium scoring, i.e. Sa36 for the EID-CT data and Qr36 kernel for the PCD-CT data. Images from both systems were reconstructed with a slice thickness of 3 mm and an increment of 1.5 mm. Details on the acquisition and reconstruction parameters are summarized in Table 1.
Table 1. CT acquisition and reconstruction parameters for energy-integrating detector CT (EID-CT) and photon counting detector CT (PCD-CT) scans.
|
EID-CT
|
PCD-CT
|
CT acquisition and reconstruction parameters
|
(SOMATOM Force; Siemens Healthineers)
|
(SOMATOM Count Plus; Siemens Healthineers).
|
Scan mode
|
Non-gated Spiral
|
Non-gated Spiral
|
CTDIvol (mGy)
|
0.88 to 1.14
|
0.92 to 1.12
|
Tube potential (kV)
|
120
|
120
|
Pitch
|
1.2
|
1.2
|
Collimation (mm)
|
192x0.6
|
144x0.4
|
Rotation time (s)
|
0.25
|
0.33
|
Monoenergetic levels (keV)
|
|
50, 65, 68, 70, 72, 150
|
Reconstruction technique
|
WFBP (standard)
|
IR1 (WFBP was not selectable)
|
Kernel
|
Sa36
|
Qr36
|
Slice thickness (mm)
|
3
|
3
|
Increment (mm)
|
1.5
|
1.5
|
Reconstruction field of view (mm)
|
160
|
160
|
Image matrix size
|
512
|
512
|
Coronary calcification inclusions
A total of 26 well-defined calcified coronary calcifications with volumes between 1 and 210 mm3 were identified and included in the study. Four to eight calcifications per heart were analysed. The calcifications were located in the left anterior descending artery circumflex artery and/or the right coronary artery
Determination and comparison of AS
The image analyses were performed by a thoracic radiologist with twenty years of radiologic experience, and approximately ten years of experience in cardiac imaging.
Evaluations of the AS were performed using the semi-automatic calcium score analysis software on a post-processing multimodality workplace (Leonardo MMWP, Siemens, Germany).
The AS values of all the included 26 calcifications were compared between the Sa36 reconstructions from the EID-CT and the monoenergetic Qr36 reconstructions from the PCD-CT. The monoenergetic levels available ranged between 50 keV and 150 keV.
Image noise measurements
Image noise was defined as the SD of the mean HU value in a 1 cm2 ROI, measured in soft tissue/myocardium in the cadaveric hearts. Two ROIs were placed in each cadaveric heart in both positions in Sa36 (EID-CT) and Qr36, energy level 72 keV (PCD-CT).
Statistics
Continuous data are presented as mean +/- SD if normally distributed, or as median and interquartile rang (IQR) if non-normally distributed. The normality assumption was checked visually using p-p-plots.
The correlation and agreement with regard to the AS, both between the two methods for each position, and between the two positions for each method, were assessed with Spearman´s rank correlation coefficient, as appropriate for non-parametric data. The agreement was investigated by means of Bland-Altman plots. Although the measurements in themselves were not normally distributed, visual assessments of p-p plots found the normality assumptions for Bland-Altman plots (differences) to hold. Statistical analyses were performed using SPSS Statistics 27 (IBM, Armonk, New York). P-values below 0.05 were considered statistically significant.