Study population
This study retrospectively enrolled 73 PA patients with various deformations who underwent preoperative DSCT and TTE examinations from June 2012 to December 2020 at the medical center of this study. The inclusion criteria were (1) patients who underwent surgery or cardiovascular angiography (CA) with PA included as the final diagnoses and (2) those who underwent both DSCT and TTE examinations before surgery or CA interventions, with the time interval between the two tests not exceeding 1 month. The exclusion criteria were (1) incomplete medical records and (2) the image quality not meeting the diagnosis requirement. Depending on the main diagnosis, all patients were divided into three groups: (1) patients with PA and ventricular septal defect (PA/VSD), (2) those with PA with VSD and major aortopulmonary collateral arteries (PA/VSD/MAPCAs), and (3) patients with PA with other major malformation (such as PA with single atrium (SA), PA with single ventricle (SV), PA with intact ventricular septum (PA/IVS), and PA with double outlet left/right ventricle).
This study was in line with the principles of the Declaration of Helsinki and approved by the Institutional Ethics Committee of West China Hospital, Sichuan University (Chengdu, Sichuan, China;No.14-163), with a waiver of informed consent due to the retrospective nature. All personal details of any patients have been removed carefully before submission.
Scanning protocol
All DSCT scans were performed on the same type of machines (Somatom Definition; Siemens Medical Solutions, Forchheim, Germany). Patients aged >5 years were trained to hold their breath during the examination. Patients aged <5 years and could not cooperate were given short-term sedation (chloral hydrate; 10% concentration; dose, 0.5 mL/kg). The scan direction was craniocaudal, and the scan range was from the inlet of the thorax to 2 cm below the diaphragm level. A nonionic contrast agent (iopamidol, 370 mg/mL; Bracco, Italy) was given intravenously (1.5 mL/kg body weight) at a rate of 1.2–2.5 mL/s, followed by 20 mL of saline solution. Automated bolus-tracking software was utilized to start the scan when the region of interest attenuation threshold reached 100 HU following a delay of 5 s. The protocol parameters (flash chest pain electrocardiogram; ECG) were as follow: tube voltage (80−120 kV; controlled by CARE KV), CARE Dose 4D (used to reduce the tube current), gantry rotation time (0.28 s), and pitch (0.2–0.5; adapted to heart rate). Moreover, the field of view was adjusted to body size.
Images were reconstructed with a slice thickness and an increment of 0.75 and 0.5 mm, respectively. The convolution kernel was I26f medium smooth ASA. Sinogram-affirmed iterative reconstruction (strength, 3) and a retrospective ECG-gated technique were used to improve image quality. All image data were transferred to a workstation (Syngo; Siemens Medical Systems, Forchheim, Germany). Several postprocessing methods (e.g., multiplanar reformations, volume rendering, and maximum intensity projection) were applied to complete image analysis (16).
Image analysis
All DSCT images were retrospectively reviewed by two veteran radiologists without knowing the surgery or DSA results. The main abnormalities were carefully reviewed at the workstation using different postprocessing methods. The diameters of the left pulmonary arterials, right pulmonary arterials, MACAPs, and descending aorta were recorded similarly to that in reference (16). The only difference in the measurement was that the main pulmonary arterial (MPA) if existing was located at the maximum of MPA. All measurements avoided narrow places. The body surface area was calculated by the Stevenson formula. The McGoon ratio, pulmonary arterials index (PAI), and total neopulmonary arterial index (TNPAI) were calculated in the standard ways.
The echocardiographic images were performed by an experienced cardiac ultrasound radiologist and checked by another senior radiologist. All detected malformations were recorded from long and short cardiac axis view at the xiphoid process, parasternum, and suprasternal fossa. The aortopulmonary window level was used to detect the collateral circulation.
Radiation dose estimation
DSCT examinations strictly adhered to the principle of “as low as reasonably achievable”. Tube voltage, tube current, and pitch were adjusted on the basis of every patient’s individual circumstance to lower the dose as much as possible as illustrated above. Volume CT dose index (CTDIvol) and dose length product (DLP) were recorded. Patients of different ages reacted differently to the radiation dose. Hence, the effective dose (ED) with different conversion coefficients was calculated as k (ED = DLP × k) on the basis of the 2007 recommendations of the International Commission on Radiological Protection(18, 19).
Statistical analysis
The Statistical Package for the Social Sciences software for Windows (version 25.0, SPSS Inc., Chicago, IL, USA) was used for statistical analysis. The Shapiro–Wilk test was used to test the normality of the variables. Bartlett’s test was used to assess the homogeneity of the variance. Categorical variables were measured as percentages and compared by Fisher’s exact or chi-square test, depending on the expected frequencies. Continuous variables were recorded as mean values or medians and were compared by one-way analysis of variance (ANOVA) followed by least-significant difference (LSD) post hoc test (normally distributed continuous variables and the variance is homogeneous) or the Kruskal–Wallis test (non-normally distributed continuous variables). The interobserver reproducibility of McGoon ratio, TPI, and TNPAI was tested by intraclass correlation coefficient (ICC) by using 32 randomly chosen patients. The ICC score of >0.75 was taken for satisfactory agreement. ANOVA followed by LSD post hoc test was used to compare subgroups of PA. Furthermore, a two-tailed p value of <0.05 was considered statistically significant in all statistical analyses.