The authors have declared that no competing interests exist.
To develop and assess a three-dimensional (3D) self-gated technique for the evaluation of myocardial infarction (MI) in mouse model without the use of external electrocardiogram (ECG) trigger and respiratory motion sensor on a 3T clinical MR system.
A 3D T1-weighted GRE sequence with stack-of-stars sampling trajectories was developed and performed on six mice with MIs that were injected with a gadolinium-based contrast agent at a 3T clinical MR system. Respiratory and cardiac self-gating signals were derived from the Cartesian mapping of the k-space center along the partition encoding direction by bandpass filtering in image domain. The data were then realigned according to the predetermined self-gating signals for the following image reconstruction. In order to accelerate the data acquisition, image reconstruction was based on compressed sensing (CS) theory by exploiting temporal sparsity of the reconstructed images. In addition, images were also reconstructed from the same realigned data by conventional regridding method for demonstrating the advantageous of the proposed reconstruction method. Furthermore, the accuracy of detecting MI by the proposed method was assessed using histological analysis as the standard reference. Linear regression and Bland-Altman analysis were used to assess the agreement between the proposed method and the histological analysis.
Compared to the conventional regridding method, the proposed CS method reconstructed images with much less streaking artifact, as well as a better contrast-to-noise ratio (CNR) between the blood and myocardium (4.1 ± 2.1 vs. 2.9 ± 1.1,
A 3D T1-weighted self-gating technique for mouse cardiac imaging was developed, which has potential for accurately evaluating MIs in mice at 3T clinical MR system without the use of external ECG trigger and respiratory motion sensor.
Recently, many kinds of animal models have become useful tools for the study of human cardiac disease processes. In particular, due to the genetic similarity with humans and the relatively low cost of maintenance, the mouse has evolved as a powerful animal model [
Routine CMR techniques for assessing MI in mouse are mainly based on external electrocardiogram (ECG) trigger and respiratory motion sensor to eliminate the cardiac and respiratory motion artifacts [
Self-gated CMR technique has potential to be an alternative tool for mouse cardiac imaging without the use of external ECG and respiratory motion sensor [
A useful strategy for accelerating data acquisition in MR imaging is to exploit the spatial or/and temporal sparsity based on the compressed sensing (CS) theory. However, previous studies only achieve 2D real-time cardiac imaging for mouse heart [
All experimental procedures were approved by the Animal Ethics Committee and conducted with strict adherence to the guidelines published by the Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences (Permit Number: SIAT-IRB-150213-YGS-ZHR-YF-A0094-3). All surgeries were performed under sodium pentobarbital anesthesia, and all efforts were made to minimize suffering.
Six C57BL/6J male mice were purchased from the Guangdong Medical Animal Experiment Center and all mice were housed in the local pathogen-free environment before experiments. Briefly, each mouse was anesthetized with sodium pentobarbital (50 mg/kg body weight) by intraperitoneal injection and maintained at 37°C. The anaesthetized mouse was intubated endotracheally in a supine position and placed on a rodent ventilator (Chengdu Taimeng Software Co., LTD, Chengdu, China). Regional myocardial ischemia was induced by transient ligation of the left anterior descending coronary artery (LAD) using a 5.0 Protune suture. Ligation was confirmed by observation of ST-elevation in a three-lead ECG and removed after 30 minutes to allow reperfusion. Finally, the chest and skin of the conducted mouse was closed with a 4.0 Protune suture and air was evacuated from the chest cavity. In addition, penicillin was injected intraperitoneally to avoid infection.
Similar to Liu’s method [
(a) The stack-of-stars sampling scheme comprises Cartesian encoding along kz direction and radial projections filling in kx-ky plane. All partition encodes for a given projection angle are collected sequentially before switching to the next projection angle. This partition-first sampling scheme enables the projection centers (denoted as black solid circles) to be used as self-gating data for extracting cardiac and respiratory motion signals; (b) Corresponding 3D GRE sequence with stack-of-stars sampling trajectories.
All MR experiments were performed on a 3 T clinical MR system (Tim Trio, Siemens, Erlangen, Germany) without using external ECG trigger and respiratory motion sensor. After a 6-hour reperfusion, a concentration of 1.5 ml/kg body weight gadolinium contrast agent (Consun Pharmaceutical Group Limited, Guangzhou, China) was diluted in 0.5 ml saline and administered via tail vein injection for the enhancement examination. The mouse was then placed head first in the supine position with the heart positioned at the center of a customized four-channel mouse coil and kept warm by a portable wax bag for MR scan. Localizations of mouse cardiac 2-chamber, 4-chamber and short-axis views were done by using a conventional 2D GRE sequence. The sequence parameters for the localization scans included: flip angle = 18°, TR = 8.6 ms, TE = 4.0 ms, field of view = 250 × 250 mm2, acquisition matrix = 200 × 256, spatial resolution = 1.3 × 1.0 mm2, slice thickness = 4.0 mm, bandwidth = 320 Hz/Pixel. After the localization was completed, the proposed 3D self-gating scan was conducted when gadolinium was injected for approximately 10 minutes to make sure MI area was enhanced. The sequence parameters for the 3D self-gating scan were: flip angle = 18°, TR = 4.2 ms, TE = 2.4 ms, field of view = 58 × 58 × 15 mm3, spatial resolution = = 0.3 × 0.3 × 1.5 mm3, partition number = 10, bandwidth = 620 Hz/Pixel. Typically, 5–6 consecutive partitions covering the entire heart from the base to the apex. Total 1600 projections were continuously collected for each mouse, corresponding to a scan time of about 1.5 minutes. Note that the spatial resolution of 3D self-gating scan was anisotropic. The spatial resolution along kz direction was 1.5 mm which was far lower than that of the kx-ky plane (0.3× 0.3 mm2). In order to match the MR images and the histology slices easily, the self-gating scan was localized based on the mouse cardiac 2-chamber, 4-chamber and short-axis views to make the kx-ky plane be in coincidence with the short axis view.
Signal processing was conducted offline with MATLAB 2013a (Mathworks Inc., Natick, MA, USA). In order to determine the cardiac and respiratory motion signals, one dimensional Fourier transform was performed on the self-gating data to derive a projection of the entire imaging volume (z intensity profile,
(a) The central points (black solid circles) of each projection angle denote a self-gating data profile. (b)
Once the respiratory and cardiac motion signals were obtained, the respiratory and cardiac phases associated with each projection angle were determined. The acquired data were then realigned according to the predetermined respiratory and cardiac phases. Specifically, the acquired data were segmented into six respiratory bins according to the respiratory positions. And then one of the six bins containing the most amounts of projections was adaptively selected for the following cardiac phase segmentation processing. The number of cardiac phases was determined based upon the mean number of projection angles in an R—R interval, and the projections were subsequently assigned to each cardiac phase for image reconstruction. To suppress the effect of arrhythmia, cardiac cycles that were 30% longer or short than the mean number were discarded.
Ideally, the data in cardiac quiescent period can be used for image reconstruction to detect MI. However, it took 42 ms for sampling a self-gating data profile in this study, leading to the self-gating data sampling rates too slow to capture the mouse cardiac quiescent period. Therefore, only the cardiac phase right before each waveform peak (near end-diastole) was used as the quiescent phase, which was similar to the process of previous study [
After the data realigned, the projection distribution in each cardiac phase was generally nonuniform and did not satisfy Nyquist-Shannon criteria, resulting in severe streaking artifacts on the reconstructed image using a conventional regridding method. To reduce these artifacts, a CS method derived by exploiting the image temporal sparsity was implemented for image reconstruction (
After MR scan, all mice were re-anaesthetized and sacrificed by rapid excision of the heart. The heart was then rinsed in 0.9% NaCl, frozen and manually sectioned into 5–6 short-axis slices with thickness of ~1.5 mm from apex to base. The slices were incubated at 37°C with 0.1% triphenyltetrazolium chloride (TTC) for 15 minutes. Both sides of each slice was photographed with a digital camera and prepared for further analysis.
To assess the accuracy of extracting cardiac and respiratory motion signal, the heart and respiratory rates were calculated from the self-gating signals, and compared with those recorded by the rodent ventilator, averaging before and after MR scans. All data are expressed as mean ± standard deviation.
To demonstrate the advantages of the proposed CS reconstruction, contrast-to-noise ratio (CNR) was calculated based upon previously published methods [
The CNR of each subject was expressed as the mean ± standard deviation. The mean CNR was statistically analyzed by the Wilcoxon sign-rank test. A
To evaluate the accuracy of the proposed method for detecting an MI, the areas of MI and myocardium were measured from the matched MR image and the histological picture, respectively. Similar to the method described in [
The boundaries (yellow solid lines) were first determined by Image-Pro Plus software and then manually outlined for reducing the measurement bias.
The mean area ratio of each subject was then obtained by averaging the areas of the selected slices in which MIs were identified. Finally, linear regression and Bland-Altman analysis were used to assess the agreement between the proposed method and the histological analysis.
All MR scans were conducted successfully. The self-gating signals characterizing cardiac and respiratory motion were successfully extracted from all subjects. As summarized in
Heart rate (bpm) | Respirtory rate (bpm) | |||
---|---|---|---|---|
Mice | Self-gating | Ventilator | Self-gating | Ventilator |
418±5 | 415±7 | 84±2 | 84±2 | |
433±6 | 429±4 | 74±2 | 78±2 | |
437±11 | 431±13 | 73±1 | 74±2 | |
430±10 | 421±4 | 76±2 | 78±2 | |
439±8 | 438±12 | 84±2 | 81±2 | |
420±4 | 420±5 | 76±1 | 79±4 | |
430±7 | 426±±7 | 78±2 | 79±2 |
A direct comparison of short-axis views of the left ventricle reconstructed by a conventional self-gated method and the proposed method is shown in
Compared with the conventional regridding method, the images reconstructed by the proposed CS method show much less streaking artifacts.
Mice | 1 | 2 | 3 | 4 | 5 | 6 | Mean |
---|---|---|---|---|---|---|---|
1.8±1.1 | 4.9±1.6 | 2.7±1.7 | 1.9±1.0 | 3.2±1.3 | 2.9±1.6 | 2.9 ± 1.1 | |
2.5±2.0 | 7.9±3.1 | 3.6±1.3 | 2.8±1.9 | 4.6±2.1 | 3.1±2.1 | 4.1 ±2.1 |
The MI regions detected by the proposed method matched those noted by histological analysis (yellow arrows in
The regions of MI (yellow arrows) matched between the histology analysis and the proposed method.
(a) Infarction and myocardial area ratio linear regression analysis of the two approaches. (b) Bland-Altman plots of the measurements. Mean difference values are represented by a solid line and the confidence intervals by dashed lines.
In this study, we developed a 3D self-gated imaging method that employs a T1-weighted GRE sequence with stack-of-stars sampling trajectories to detect MI in a mouse model at a 3T clinical MR system. Our preliminary
There are several important aspects that assure the success of the proposed method for detecting MI in the mouse. First, a large flip angle of 18o was used in the GRE sequence to facilitate heavy T1-weighted imaging. It has previously been reported that post-contrast, T1-weighted imaging techniques are sensitive for the detection and delineation of MI [
Compared with the conventional 3D self-gated method employed in the mouse [
The self-gated method presented here was performed at 3T clinical MR system, which may be valuable for such an animal study. Due to the small size of the mouse, imaging of mouse cardiac MI is typically performed at high-field (≥ 4.7 T) animal MR system for achieving high spatial resolution images [
This study had two limitations. First, off-line reconstruction remains a problem for routine application. Although the stack-of-stars sampling pattern helps to enable parallelized slice-by-slice reconstruction, the CS reconstruction remains time-consuming in its current version. This issue may be addressed using parallel GPU [
In conclusion, a 3D self-gated cardiac imaging technique, using a T1-weighted stack-of-stars GRE sequence and CS, was developed for the evaluation of MI in mouse model. A preliminary