† These authors contributed equally to this work.
‡ Corresponding author. E-mail:
Project supported by the National Natural Science Foundation of China (Grant Nos. 81501463, 61471349, 81671853, 81571669, and 61671026), the National High Technology Research and Development Program of China (Grant No. 2015AA043203), the Natural Science Foundation of Beijing, China (Grant No. 7162112), Guangdong Innovative Research Team Program of China (Grant No. 2011S013), the Natural Science Foundation of Guangdong Province, China (Grant Nos. 2014A030310360 and 2014A0202015028), the Beijing Center for Mathematics and Information Interdisciplinary Sciences, Shenzhen Fundamental Research Program, China (Grant Nos. JCYJ201500731154850923 and JCYJ20140417113430665), Shenzhen High-level Oversea Talent Program, China (Grant No. KQJSCX20160301144248), and the Nanshan Technology Research Fund, China (Grant No. KC2014JSQN0001A).
Conventional multiple breath-hold two-dimensional (2D) balanced steady-state free precession (SSFP) presents many difficulties in cardiac cine magnetic resonance imaging (MRI). Recently, a self-gated free-breathing three-dimensional (3D) SSFP technique has been proposed as an alternative in many studies. However, the accuracy and effectiveness of self-gating signals have been barely studied before. Since self-gating signals are crucially important in image reconstruction, a systematic study of self-gating signals and comparison with external monitored signals are needed.
Previously developed self-gated free-breathing 3D SSFP techniques are used on twenty-eight healthy volunteers. Both electrocardiographic (ECG) and respiratory bellow signals are also acquired during the scan as external signals. Self-gating signal and external signal are compared by trigger and gating window. Gating window is proposed to evaluate the accuracy and effectiveness of respiratory self-gating signal. Relative deviation of the trigger and root-mean-square-deviation of the cycle duration are calculated. A two-tailed paired t-test is used to identify the difference between self-gating and external signals. A Wilcoxon signed rank test is used to identify the difference between peak and valley self-gating triggers.
The results demonstrate an excellent correlation (P = 0, R > 0.99) between self-gating and external triggers. Wilcoxon signed rank test shows that there is no significant difference between peak and valley self-gating triggers for both cardiac (H = 0, P > 0.10) and respiratory (H = 0, P > 0.44) motions. The difference between self-gating and externally monitored signals is not significant (two-tailed paired-sample t-test: H = 0, P > 0.90).
The self-gating signals could demonstrate cardiac and respiratory motion accurately and effectively as ECG and respiratory bellow. The difference between the two methods is not significant and can be explained. Furthermore, few ECG trigger errors appear in some subjects while these errors are not found in self-gating signals.
Cardiac cine magnetic resonance imaging (MRI) has been widely used to non-invasively assess cardiac morphology, function and blood flow.[1–3] Conventional cardiac cine MRI is typically based on the segmented data acquired by multiple breath-holds with two-dimensional (2D) or three-dimensional (3D) balanced steady-state free precession (SSFP) pulse sequences.[4–8] In such acquisitions, the breath-hold positions monitored by respiratory bellow (RB) or air bag may be inconsistent, which leads to slice misregistrations or motion artifacts.[9] Furthermore, the RB signal may not accurately reflect respiratory motion if the bellow is not installed at the optimal location. The breath-hold 3D SSFP technique may eliminate slice misregistration limitation and improve the signal-to-noise ratio (SNR).[7,8] However, the long breath-holding time (15 s–20 s) may be difficult for some patients, especially pediatric patients or patients with congestive heart failure.[10] Furthermore, both 2D and 3D SSFP techniques are characterized by limited spatial resolution.[11–13] The conventional 2D breath-hold SSFP technique also requires electrocardiographic (ECG) gating as the external trigger to synchronize cardiac motion.[6–8] However, in some cases, ECG triggers are unreliable because of the influence of the magnetic field of the MRI system or the physiological abnormalities of the patient.[9,14] Inaccurate ECG triggers will compromise the accuracy of k-space data of each cardiac phase. Therefore, triggering errors constitutes an additional disadvantage in ECG gated cardiac cine MRI.[9]
Cardiac and respiratory self-gated free-breathing SSFP has recently been proposed as an alternative to conventional breath-hold SSFP for cardiac cine MRI.[15–18] Free-breathing acquisition enables better spatial resolution. Self-gating signals derived from the k-space data can theoretically and technically replace the external gating signals. There are many methods to obtain self-gating signals, such as k-space center data points,[9] k-space lines across the k-space center,[17] or low-resolution images.[19,20] However, the accuracy and effectiveness of self-gating signals have been barely systematically studied.
In this study, external monitored signals are simultaneously acquired by ECG and RB devices during the free-breathing 3D SSFP acquisition. Self-gating signals are derived via repeated acquisition of the k-space centers along the z-direction.[16] The accuracy and effectiveness of the self-gating signals are evaluated by comparing with external monitored signals.
A previously described self-gated free-breathing 3D SSFP technique[16] was used. Twenty-eight healthy volunteers (16 males and 12 females), whose ages ranged from 25 to 35 years (median age 30.5 years), were included in this study. The self-gated free-breathing 3D SSFP technique was performed on a 1.5-T GE HDx scanner (maximum gradient amplitude 33 mT/m, slew-rate 120 (mT/m)/s, Excite 14M5 platform; GE Healthcare, Waukesha, WI, USA). The study was approved by the local institutional review board at Weill Cornell Medical College, and written informed consent was obtained from all subjects prior to enrollment. An eight-channel cardiac phased-array coil was used for signal collection. The cine imaging parameters were set as follows: TR/TE = 3.5/1.3 ms, flip angle = 40°, bandwidth = ± 125 kHz, slice thickness = 7 mm (no gap), number of slices = 12–14, number of profiles = 5000, and total acquisition time ≈ 5 min.
The heart/breath rate and the cardiac/respiratory cycle duration were compared between the self-gating and ECG/RB methods. Trigger positions were evaluated by calculating the correlation coefficients and trigger deviation time of the two methods. The triggers of the self-gating signals were determined by the signal peaks and valleys. A Wilcoxon signed rank test with a two-sided alpha of 0.05 was used to identify differences between peak and valley triggers. There were individual differences in the heartbeat and breath patterns among the healthy volunteers; thus, the relative deviation between the self-gating and external triggers was calculated via the following formula:
Figure
Figure
All subjects are successfully acquired with the free-breathing self-gated 3D SSFP sequence. Twenty-five cases are included in the data analysis because three cases failed to provide effective externally monitored signals or triggers for comparison. The cardiac cycle duration (R–R intervals) and the respiratory cycle duration of 25 subjects are 751.3±93.1 ms (which corresponds to 81.1±10.3 beats per minute) and 3314.7±1072.6 ms (which corresponds to 19.5±4.5 breaths per minute), respectively.
Figure
The cardiac and respiratory signal comparisons between the two methods are summarized in Tables
![]() | Table 1. Cardiac motion comparison between the self-gating and externally monitored signals. . |
![]() | Table 2. Respiratory motion comparison between the self-gating and externally monitored signals. . |
Figure
![]() | Fig. 5. Statistical results of the ratios of the RMSD to cycle duration. Triggers from peaks and valleys of the self-gating signals are compared. |
Figure
Figure
![]() | Fig. 7. An inaccurate ECG trigger (indicated by the black arrow) is recorded in a representative case. The amplitudes of all curves are rescaled for comparison only. |
Conventional cardiac cine MRI is typically performed by using multiple breath-hold techniques. The newly proposed self-gated free-breathing 3D SSFP technique enables the completion of the scan with neither peripheral monitors nor breath-holds. In this study, the accuracies and effectiveness of the self-gating signals are systematically investigated by comparing the self-gating signals with the synchronized externally monitored signals.
Self-gating signals directly reflect the cardiac and respiratory motion compared with the externally monitored signals. The two-sided Wilcoxon signed-rank test shows that there is no significant difference between peak and valley triggers at the 5% significance level for self-gating signals. However, the relative deviation and statistical analysis of RMSD show that the peak CSG and valley RSG triggers enable better synchronization with the corresponding external triggers. For cardiac motion, an ECG signal reflects the cardiac electric activity,[21,22] and ECG triggers are recorded in atrial systoles. The CSG signal reflects the change in heart blood volume.[17] Thus, the MR signal intensity increases during diastole. The peaks of self-gating signals correspond to end-diastoles. Therefore, the differences between peak triggers and ECG triggers are smaller, and the trigger deviation time of peak triggers is shorter. For respiratory motion, RB triggers are recorded in the early phase of expiration. Valleys of the RSG signal correspond to end-inspiration. Furthermore, as a result of the respiratory signal fluctuation at the end of expiration for some volunteers, an accurate detection of peak is difficult. Therefore, the difference between the valley trigger and RB trigger is smaller, and the absolute trigger deviation time of the valley trigger is shorter. Note that a negative deviation time denotes the external trigger occurs prior to its corresponding self-gating trigger. Furthermore, the temporal resolutions of the self-gating signal and monitored signal are different.
Conventional cardiac cine MRI typically requires ECG triggers to synchronize the k-space data acquisition with cardiac motion. However, rapid switching of the magnetic field gradients will compromise the accuracy of ECG trigger because of the magnetohydrodynamic effect. Patients with abnormalities in cardiovascular structures or chest geometry may exhibit low-amplitude ECG signals that make it difficult to implement the trigger selection.[9] The inaccurately recorded ECG triggers (shown in Fig.
In this study, a novel method is introduced to evaluate the RSG signals. In the free-breathing acquisition, the acquired k-space data within the given gating windows are used for image reconstruction.[16] The gating windows need to balance the undersampling artifacts (narrower gating windows) and motion blurring (larger gating windows).[16,17] Therefore, the similarity between the RSG and RB signals compared by the locations of these gating windows is proposed to compare the respiratory signals. The percentages of data located in both the RSG and RB gating windows are calculated. Figure
In this study, all subjects are healthy volunteers who are assumed to have a regular heartbeat and respiration. However, some patients, especially patients who suffer from cardiac arrhythmias, have irregular heartbeats.[23–27] Therefore, the accuracy and effectiveness of self-gating signals in these subjects requires further investigation. In addition, the k-space sampling scheme of data acquisition adopted in this study is modified from a 3D golden-ratio based radial sampling scheme. However, there are many other methods used to obtain self-gating signals, such as the respiratory self-gated multiple-gradient recalled echo sequence[28,29] and manifold learning-based 2D golden-ratio based radial SSFP technique.[30] Therefore, the accuracy and effectiveness of self-gating signals obtained with other sampling schemes and comparisons between them also require further investigation.
In conclusion, there is no statistically significant difference between self-gated free-breathing SSFP and conventional ECG/RB techniques. Self-gating signals synchronize well with externally monitored signals and provide approximately the same cardiac cycle duration and respiratory cycle duration compared with the monitored signals. There is no significant difference between peak and valley triggers for both cardiac and respiratory motions. However, peak CSG triggers and valley RSG triggers enable better synchronization with the corresponding external triggers and are more effective. Furthermore, self-gating triggers are more reliable than external triggers in free-breathing imaging.
1 | |
2 | |
3 | |
4 | |
5 | |
6 | |
7 | |
8 | |
9 | |
10 | |
11 | |
12 | |
13 | |
14 | |
15 | |
16 | |
17 | |
18 | |
19 | |
20 | |
21 | |
22 | |
23 | |
24 | |
25 | |
26 | |
27 | |
28 | |
29 | |
30 |