1.
Patients and study design are described in table 1
Study design
|
Prospective
|
Ethics Committee
|
Approved
|
Written informed consent
|
All patients
|
Inclusion criteria
|
CCTA referral
|
Exclusion criteria
|
- < 18 years
- Contraindications to X-ray exposure,
Allergy to iodinated contrast agents
- Mild to severe renal insufficiency (creatinine clearance < 60ml/min/1.73m² body surface area)
|
Inclusion period
|
June - July 2013
|
Table 1: Study design and cohort
2.
Coronary CT acquisition protocol and image processing.
All patients underwent the same imaging protocol using a 2nd generation CT scanner (Aquilion ONE ViSION Edition; Toshiba Medical Systems,
Otawara,
Japan) equipped with 320 rows and a gantry rotation time of 275 ms.
The protocol included a coronary calcium score scan (120kVp; 300mA; 3 mm slice thickness) and contrast-enhanced CCTA using the parameters described in table 2.
Scanner |
Aquilion ONE ViSION Edition; Toshiba Medical Systems,
Otawara,
Japan |
N° of rows |
320 |
Maximum gantry rotation time |
275 ms |
Detector configuration; single step z-axis coverage |
320 x 0.5 mm; 160 mm |
Scanning mode |
Axial,
ECG-Triggered,
N° of rotation depending on the heartrate
· 1 rotation HR < 75 bpm
· 2 rotations 75 bpm < HR < 100 bpm Arrhythmia detection
|
Patient preparation |
0.4 mg of sublingual nitroglycerine 2 min before scanning |
Tube voltage/current |
Automatic based on scout views (SUREexposure3D,
Toshiba Medical Systems). |
Contrast injection protocol |
- Dual injection of 100% Iomeron 400 (Bracco Diagnostics,
Milan,
Italy; 400 mg of iodine/ml) for 9s (+ 1s for every supplementary tube rotation) + 30 ml saline
- Injection rate adapted to the automatically set tube voltage/current:
- 3.5 ml/sec for 80 kVp protocol
- 4 ml/sec for 100 kVp and < 450 mA
- 5 ml/sec for 100 kVp and > 450 mA
- 6 ml/sec for 120 kVp protocol.
|
Scan triggering |
Automated bolus tracking |
Slice thickness/increments |
0.5/0.25 mm |
Image reconstruction matrix |
512² |
Reconstruction,
algorithms and kernel |
Asymmetric cone beam reconstruction AIDR3D (Toshiba Medical Systems) Kernel FC03 (Standard) Best diastolic phase ± increments of 3%,
in case of movements,
depending on the heartrate. |
Table 2: Imaging consisted of calcium scoring and CCTA protocols.
3.
Analysis of image quality
Images were evaluated by 2 independent experienced readers on a dedicated workstation equipped with software enabling automatic centerline determination of each of the coronary segments.
The window level and width levels were set to 300 and 800 respectively.
They evaluated independently motion-related image quality on longitudinal and orthogonal reconstructions for all coronary segments ≥ 1.5 mm in diameter according to the American Heart Association segmentation (S1-S15) by using a four-point Likert scale (Table 3).
Grade
|
|
1
|
no artifact
|
2
|
minor,
mild artifact
|
3
|
moderate artifact but still interpretable
|
4
|
Severe artifacts rendering diagnostic interpretation not possible
|
Table 3: 4-point Likert scale for motion-related image quality
Consensus readings were performed to resolve all discordant inclusion criteria and segment location.
Image noise was evaluated by assessing the noise levels in the ascending aorta and the endobronchial air (Hounsfield unit and standard deviation).
Signal to noise ratio (SNR) is defined as the ratio between the mean attenuation and the standard deviation in the ascending aorta.
4. Statistical analysis
All patient and examination data were stored in excel sheet.
Interobserver agreement for image quality was calculated using Cohen’s κ test by using the following scale: κ values of less than 0.20 are indicative of poor agreement; 0.21-0.40,
fair agreement; 041-0.60,
moderate agreement; 0.61-0.80,
good agreement; 0.81-1.00,
excellent agreement.
The student t test was used to compare continuous variables,
and chi-square test was used to compare nonparametric variables.
A p-value of less than 0.05 is considered to express a statistically significant.
To identify the predictors of the radiation dose (mSv),
a linear regression model was used with continuous variable mSv as outcome and the following variables as potential predictors: age,
beta-blocker administration,
the number of heart beats,
presence of arrhythmia,
scanning length,
Agatston calcification score,
heart rate,
BMI and SNR.
A general linear model was fitted with a backwards model selection.
The model assumptions of normality,
constant variance and linearity were checked.
For identifying the factors with impact on image quality,
the following potential risk factors are investigated using a generalized linear mixed model (GLMM): diameter of the vessel,
the age of the patient,
use of beta-blockers,
the number of heart beats (1,
2 or 3),
arrhythmia,
scan length,
Agatston calcification score,
SNR,
heart rate,
radiation dose and BMI. The original outcome for the mean quality of the two readers is a multinomial variable with an ordinal scale (grade 1,
1.5,
2 and 3).