Abstract
BACKGROUND:
Accuracy of CT-derived fractional flow reserve depends on good image quality. Thus, improving image quality during coronary CT angiography (CCTA) is important.
OBJECTIVE:
To investigate impact of respiratory motion artifact on coronary image quality focusing on vessel diameter and territory during one beat CCTA by a 256-row detector.
METHODS:
We retrospectively reviewed patients who underwent CCTA under free-breathing (n = 100) and breath-holding (n = 100), respectively. Coronary image quality is defined as 4-1 from excellent to poor (non-diagnostic) and respiratory motion artifact severity is also scored on a 4-point scale from no artifact to severe artifact. Coronary image quality and respiratory motion artifact severity of all images were evaluated by two radiologists independently.
RESULTS:
Compared with free-breathing group, the image qualities are significantly higher in per-segment, per-vessel and per-patient levels (P < 0.001) and proportion of segments with excellent image quality also improves significantly (73.6% vs 60.1%, P < 0.001) in breath-holding group. The image quality improvement occurs in medium-sized coronary arterial segments. Coronary image quality improves with respiratory motion artifacts decreasing in both groups, respectively.
CONCLUSION:
During one heartbeat CCTA, breath-holding is still recommended to improve coronary image quality due to improvement of the image quality in the medium-sized coronary arteries.
Introduction
Coronary computed tomographic angiography (CCTA) is gaining widespread acceptance for noninvasive evaluation of coronary arteries [1–3], and its clinical use has been recognized by the latest guidelines on the stable coronary artery disease management [4]. Two of key patient-related factors for successful CCTA are: patient cooperation in breathing and heart rate (HR) control [5]. Recently, several technologies have been introduced in clinical practice for improving image quality, including CT machines with dual sources or wider detector coverages [6–8]. Due to wide volume coverage, some CT machines are able to perform an electrocardiogram (ECG)-gated examination of the heart during one heartbeat without movement of the table [9].
In recent years, the CT-derived fractional flow reserve (FFRCT) has developed rapidly. Given the FFRCT reliance on accurate coronary and myocardial segmentation for proper image-based modeling, excellent image quality should remain a primary goal during CCTA [10, 11]. Theoretically, in the case of CCTA using the 256-slice scanner during one cardiac cycle, gentle steady breathing may not produce an additional significant motion artifact [12]. Previous studies have showed that CCTA using the 256-detector row CT can be performed in a single cardiac cycle with high diagnostic value without a breath-holding instruction [6]. Based on the above results, free-breathing data acquisition was advocated due to simple process management. However, the diagnostic evaluability of free-breathing CCTA indeed decreased on a per-segment level [7]. Furthermore, the potential reasons and the contribution of respiratory motion artifact to this decrease are still not clear.
Establishment and strict adherence to same CCTA imaging protocols with appropriate training is a viable way of improving the quality of imaging and improve the FFRCT accuracy [13]. Improved subjective image quality can offer better the diagnostic performance of FFRCT [14]. To date, free-breathing and breath-hold protocols are both used during the one-hart beat CCTA imaging. Breath-holding technique should be preferred if breathing-hold could improve the subjective image quality. Thus, the aim of the present study was to evaluate the respiratory motion artifact effect on coronary arterial image quality focusing on vessel diameter and territory by comparing the image quality between breath-holding and free-breathing CCTA performed with a 256-row detector CT.
Materials and methods
Patients
This retrospective study was approved by the institutional review board and written informed consent was obtained from all individual participants included in the study. One hundred and five patients who underwent free-breathing CCTA (group one, free-breathing group) and another 103 patients who underwent end-inspiration breathing-holding CCTA (group two, breath-holding group) from July 2017 to November 2019 were retrospectively included in the study. CCTA was used to detect the coronary stenosis because most of the including patients suffered from chest pain. Exclusion criteria were allergy to iodine-containing contrast medium (n = 4), renal insufficiency(n = 2), uncontrolled hyperthyroidism (n = 1), and coronary arterial bypass surgery (n = 1). At last, one hundred patients in free-breathing group and another 100 patients in breath-holding group were retrospectively enrolled. In the enrolled patients, an invasive coronary angiography (ICA) were selectively performed based on the results of CCTA in some patients. ICA was applied as the gold standard for diagnostic accuracy of coronary stenosis by CCTA if the interval between ICA after CCTA within 1 month.
CCTA protocol
All patients were scanned using a 256-detector row CT (Revolution CT, GE Healthcare, Waukesha, WI USA). The maximal coverage of the wide detector in the z-axis was 16 cm, to provide a scan range from the tracheal bifurcation to the bottom of the heart. Prospectively ECG-triggered CCTA with axial acquisition was performed within a single cardiac cycle. Automatically selected tube voltage was set by kV Assist and tube current by Smart-mA based on the scout image of the patients. The scanning time was within 0.3–1.0 s. No additional beta blocker was administered immediately prior to the CT scans. Non-ionic iodine-contained contrast media (Ultravist, 370 mg I/mL; Bayer Health Care, Berlin, Germany) was injected at a rate of 4–5.5 ml/s via the median cubital vein, at a dose of 0.8 ml/kg of body weight [6]. The CCTA protocols and contrast media delivery protocols were exactly the same in the two groups except for the breathing status. One was the free-breathing group and the other was the end-inspiration breath-holding group after short and concise practicing. The snapshot freeze technique was used to correct the motion artifacts [15].
Image analysis
CCTA images were transferred to a GE Advanced Workstation (AW4.6) for post-processing and analysis, with analysis performed with volume rendering, curved planar reconstruction, and multiple planar reconstructions.
Coronary artery calcium was quantified by Agatston score with GE Advanced Workstation (AW4.6) and classified into no, mild, moderate and severe categories (0, 1 to 99, 100 to 399, ≥400) [16, 17]. Attenuations of aortic root immediately cranial to the left main coronary artery (LM), LM and proximal segment of right coronary artery (RCA) were obtained by placing a region of interest (ROI) in the center of the target arteries with nearly 75% of its lumen area avoiding atherosclerotic plaques. Image noise was defined as the standard deviation of CT density in the aortic root [1]. The signal-to-noise ratio is obtained by dividing the mean density of the aortic root by the noise. To determine the image quality, the 18-segment coronary artery tree model was used according to the guidelines proposed by the Society of Cardiovascular Computed Tomography [7]. All segments <1.5 mm in diameter were excluded from evaluation. A 4-point Likert scale was used to qualitatively assess the coronary image: 4, excellent image quality free of artifacts; 3, good image quality with minor artifacts, but fully evaluable and diagnostic; 2, adequate image quality with moderate artifacts, but acceptable for diagnosis; 1, poor image quality with severe artifacts and non-diagnostic [18]. Examples of coronary image quality scores from 4-1 are shown in Fig. 1. Similarly, the respiratory motion artifact severity was scored on a 4-point scale: 4, no artifact; 3, mild artifact; 2, moderate artifact; and 1, severe artifact [19, 20] and examples are shown in Fig. 2. Moreover, the coronary segments with diameter stenosis ≥50% were separately recorded by CCTA and ICA. All the above image analysis was performed independently by two radiologists (both with over 10 years of clinical experience in CCTA performance) who were blinded to breath-hold status and clinical information.
Representative images demonstrate the different image quality scores of the coronary arteries. (A: 4 = excellent, B: 3 = good, C: 2 = adequate, and D: 1 = poor).
Representative images demonstrate the different respiratory motion artifact scores. (A: 4, no artifact; B: 3, mild; C: 2, moderate, and D: 1, severe).
Statistical analysis was performed using the SPSS 17.0 (SPPS Inc., Chicago, Ill, USA). The normality of data distributions was analyzed using the Kolmogorov–Smirnov test. All numeric data with normal distributions were reported as mean±SD, and an independent sample t test was adopted. Otherwise medians (25th-75th percentile) were reported, and the Mann-Whitney U test was used. The observer agreement of image quality scoring was tested by the Cohen’s kappa. This was interpreted as moderate for 0.40 <kappa <0.60, good for 0.60 <kappa <0.80, and excellent for kappa >0.80. The diagnostic accuracy of CCTA to detect ≥50% diameter stenosis defined by ICA was calculated from the Chi-square test of the contingency table on the per-segment level. Patient characteristics, scan-related parameters, image quality of coronary arteries, and respiratory motion artifacts of the two groups were compared using Mann-Whitney U test and t-tests if applicable. P-values <0.05 were considered statistically significant.
Results
Patient characteristic and scan-related parameters
Patient characteristic comparisons including age, sex, body weight, height, and heart rate, indicated no statistically significant differences between the two groups. Compared with the free-breathing group, the reconstruction phase (percentage of the RR interval) during R-R interval was a little earlier in the breath-holding group (free-breathing group one 61.3% vs. breath-holding group 54.3%, P = 0.002). Table 1 demonstrates detailed information of patient and scan-related characteristics. The coronary calcium was a little higher in the free-breathing group (p = 0.043). There were no significant differences in the distributions of Agatston scoring categories between the two groups (P = 0.085).
Patient demographic data and scan-related parameters
Patient demographic data and scan-related parameters
Numeric data with normal distributions were reported as mean±standard deviation and otherwise medians (25th-75th percentile) were reported. *Calcium score was non-normally distributed data. There were 35 patients with coronary calcification in free breathing group and 21 in the breath holding group. Mann-Whitney U test indicated that calcium score was higher in the free breathing group.
No significant differences were detected between the groups in attenuations obtained in the aortic root (group one 464.4±73.9 vs. group two 454.4±74.1 HU, P = 0.342), left main coronary artery (group one 438.1±75.3 vs. group two 422.7±71.9 HU, P = 0.112) or proximal segment of right coronary artery (group one 426.3±73.2 vs. group two 408.9±72.2 HU, P = 0.092). The mean image noise in the aortic root in the breath-holding group was higher than that in the free-breathing group (breath-holding group 29.8±5.3 vs. free-breathing group 25.5±3.9 HU, P < 0.001), resulting in a lower signal-to-noise ratio (breath-holding group 15.6±2.8 vs. free-breathing group 18.4±2.7, P < 0.001).
A total of 1,222 segments were included for evaluation in the breath-holding group and 1,220 segments in the free-breathing group as well. Intrareader reproducibility of image quality evaluation between the two independent readers was excellent (kappa = 0.81). The image qualities in the breath-holding group were statistically higher than the free-breathing group in all segment-, vessel-, and patient-based levels (P < 0.001). Compared with the breath-holding group, there was a higher percentage of segments that could not be assessed in the free-breathing group (1.2% vs. 0.2%, P = 0.002). Detailed image quality score of the two groups are shown in Table 2.
Comparison of coronary image quality between the two groups based on patient-, vessel-, and segment-level
Comparison of coronary image quality between the two groups based on patient-, vessel-, and segment-level
In the breath-holding group, mean image quality score improvements were achieved in the RCA, left anterior descending (LAD), and left circumflex (LCX) arteries, respectively. Further, the breath-holding group demonstrated an incremental proportion of RCA, LAD, and LCX with excellent image quality (score 1), which is shown in Fig. 3. Different coronary sizes showed different impacts on the image quality improvement by comparing the image quality in both groups. Image qualities were mainly improved in the medium-sized coronary arteries including the middle and distal segments of RCA, LAD, and LCX as well as their proximal branches. Not only segments with relatively larger size (LM and the proximal segments of RCA, LAD, and LCX), but also segments with relatively smaller size (second diagonal and second obtuse marginal branches) demonstrated no image quality improvement or decreasing in the breath-holding group.
An incremental proportion of the right coronary artery (RCA), left anterior descending (LAD), and left circumflex (LCX) with excellent image quality is respectively demonstrated in breath-holding group when comparing with the free-breathing group.
Obviously, the breath-holding group demonstrated fewer respiratory artifacts than the free-breathing group, which is shown in Table 3. Figure 4 shows that the coronary image qualities improved with respiratory motion artifacts decreasing in both groups. We divided the patients into two subgroups with (score 1–3) or without (score 4) respiratory motion artifacts according to respiratory motion score. In the subgroup without respiratory motion artifacts, the image qualities of the breath-holding group were also improved than the free-breathing group (P = 0.009). Table 4 demonstrates the image quality comparison between the two subgroups based on respiratory artifacts.
Comparison of respiratory motion artifacts between the two groups
Comparison of respiratory motion artifacts between the two groups
The mean coronary image qualities improved with respiratory motion artifacts decreasing in both groups.
Comparison of mean coronary image quality score in terms of respiratory motion artifact
Seven patients in the free-breathing group and eight in the breath-holding group underwent ICA after CCTA exam within one month. The diagnostic accuracy of CCTA (≥50% diameter stenosis defined by ICA) on per-segment level between the two groups presented no statistically significant differences (93.7% vs. 93.9%).
Discussion
Our study found that the respiration movement artifacts still affect image quality by single heartbeat imaging technology during CCTA. With a simple breath-holding instruction, the proportion of coronary arteries with excellent image quality were all improved in RCA, LAD, and LCX, mainly contributed by the image quality improvement in the medium-sized coronary segments (middle and distal segments of RCA, LAD, and LCX as well as their proximal branches).
As a common understanding in the literature, CCTA examinations should be performed under conditions that minimize motion artifacts to improve the diagnostic accuracy for coronary artery disease [21]. During the free-breathing state, the heart position is significantly different between the two or three cardiac cycles, especially among patients with very fast or irregular heart rates [22]. In our study, although free-breathing CCTA has similar higher diagnostic accuracy for detecting coronary artery stenosis, holding their breath-holding yields higher image quality. Further, many prior studies of FFRCT studies exclude a non-negligible proportion of coronary segments from analysis because of impaired image quality [23]. Careful adherence to proper image acquisition and future improvements in CT technology will be undoubtedly helpful to reduce this exclusion [10]. In our study, we demonstrated that the proportions of coronary arteries with excellent image quality were all increased following a simple breath-holding instruction, which is good for the FFRCT evaluation. However, we also found that breath-holding can alleviate but not totally eliminate the respiratory motion artifact. This can be mostly explained by the point that some patients cannot fully follow the breath-holding instruction. Furthermore, the research of Katsuda et al. [24] shows that a major respiratory motion artifact source is diaphragmatic excursion, and diaphragm motion was largely suppressed until 7 s later. In our study, we began our data acquisition at 4–5 s after the breath-holding instruction.
The main factors determining the image quality of coronary arteries include the vessel size and cardiac and respiratory motions. Breath-holding is an effective way to suppress the respiratory motion. In our study, breath hold improves the overall image quality mainly contributed by the improvement in the middle coronary artery segments and their large branches. Moreover, we found that the improvement in interpretability of some segments was not statistically significant, including some large segments such as the LM and proximal segments of RCA, LAD, and LCX because the image quality of these segments was relatively very good no matter the breathing status. No improvement in image quality of these large segments in breath-hold group was also partly explained by that these large segments are stabilized by the great vessels. Non-improvement in the image quality of the small branches by breath-holding method is possibly explained by that the image quality of small size coronary arteries are mainly influenced by their size and cardiac motion.
Image noise and signal noise ratio are commonly used objective parameters to evaluate the image quality [25]. The lower signal noise ratio may lead to a lower subjective image quality score. In our study, the image noise was a little higher in the breath-holding group, which lead to a reduced signal noise ratio. But the subjective image quality score was higher than that in the free-breathing group. This result indicate that the breath-hold could improve the image quality although the signal noise ratio decreases slightly. Coronary calcium score in the free-breathing group was a little higher than the breath-holding group, but the Agatston scoring categories was similar. So, the impact of coronary calcium on the image quality could regard as no difference in the two group.
There were some limitations to this study. First, we compare the image quality of the different patient groups, but not the same patient group performed by two different breathing statuses. Second, the sample size of our study was relatively small, which may affect the statistical results. Third, the image quality was quantitatively accessed, not quantitatively analyzed, such as the method explained by Gordic et al. [26] Fourth, the number of patients with ICA was very small; hence, the diagnostic accuracy comparison may not be propagable.
In conclusion, during one heartbeat CCTA imaging, breath-holding is still preferred to improve the subjective image quality in all three coronary arteries mainly due to the improvement of the medium-sized coronary arteries.
Conflict of interest
This research was partly supported by the National Natural Science Foundation under grant of 81701651. We have full control of all primary data and agree to allow the journal to review the data if requested.
Footnotes
Acknowledgments
This work was done in department of radiology, the first affiliated hospital of Nanjing medical university. The authors thank Yongyue Wei, PhD from the Nanjing medical university for support on the statistical analysis of this manuscript.