Background

The clinical evaluations of cardiac function in CMR relies on breath-held ECG-gated cine images. However, the traditional method suffers from inefficiency, necessitating 10-12 breath-holds to cover the whole left ventricle (LV), and is prone to the introduction of respiratory-motion and ECG-gating artifacts, especially at 3T. In a previous study, we presented an effective 2D Spiral Acquisition with Respiratory Correction and Cardiac Self-gating (SPARCS) technique [1]. To further improve the scan efficiency, we have now developed an innovative CAIPIRINHA-based Simultaneous Multi-Slice (SMS) SPARCS technique.

Methods

The SMS pattern was developed based on CAIPIRINHA principles [2]. 4 different phase modulation strategies were evaluated. Firstly, the Hadamard encoding was utilized to design a 4 spiral interleaves pattern with multi-band (MB) factor of 3, as shown in Figure 1(a). As Hadamard encoding only allows for an even number interleaf pattern, we further modified the pattern as follows. The phase modulation φ of nl (nl = 1, 2, … NL) interleaf at ns slice with MB factor is:$$\varphi_{nl, ns, MB}=\frac{2 \pi(ns-1)(nl-1)}{MB} \quad[\text { Eq. } 1]$$ To make sure the sum of all NL interleaves can be divided by 2π, the sum of all the interleaves equals to $$$\frac{\pi(NL-1)NL}{MB}$$$. The normalization factor can be calculated as $$$\frac{2MB}{(NL-1)NL}$$$. With a 40 ms temporal resolution for Cine, we designed the phase modulation pattern based on Eq.1, normalization factor, 5 interleaves and MB = 3, as shown in Figure 1(b). Then, the normalization factor was doubled to generate the 3rd pattern in Figure 1(c), and this design allows for more spreading-out distribution of the spiral trajectories in the FOV. Lastly, we designed the 4th pattern (Figure 1(d)) by keeping the phase of the first two slices the same as the 3rd pattern, but reversing the phase of the 2nd slice and assigning it to the 3rd slice. The 4 different phase pattern designs were demonstrated in an ACR phantom.

Since the acquisition scheme is continuous and the cardiac motion state is extracted using a self-gating strategy, to avoid the frequency disruption brought by sampling pattern, simulation and in-vivo experiment were both performed to test the rotating angle among every 5 spirals, with golden angle (GA) equals to 137.5º and 49.8º[3] (Figure 2).

As shown in Figure 3, self-gating cardiac triggers were extracted from sliding-window heart image navigators, facilitating real-time tracking of cardiac motion. Utilizing the automatically detected ROI around the heart, images from each heartbeat underwent rigid registration for respiratory motion correction. Subsequently, the registered k-space data were retrospectively segmented into different cardiac phases. Images from distinct slices were reconstructed using the SMS-Slice-L1-SPIRiT method [4].

11 volunteers were imaged on a 3T Siemens scanner. The SMS-SPARCS data was acquired continuously for 10 seconds per slice group using spoiled-GRE-based spiral trajectories under free-breathing and no ECG-gating. To achieve comprehensive coverage of the LV, each subject typically underwent 3-4 slice groups. Conventional breath-hold ECG-gated bSSFP cine images were acquired for comparison. The results from both the proposed technique and the clinical standard were evaluated blindly, with grading assigned on a scale of 1 (poor) to 5 (excellent) by 2 experienced cardiovascular imagers. Additionally, LV ejection fraction (LVEF) was calculated and compared using Bland-Altman analysis.

Results

A comparison of the 4 phase modulation patterns is shown in Figure 1, where (c) and (d) show better image quality than (a) and (b). To make the design more generalizable for different interleaves and MB factors, the phase pattern in Figure 1(d) was utilized for later in-vivo experiments. From the simulation part in Figure 2, there is a clear frequency disruption of GA 137.5º in the cardiac motion frequency range (0.5 to 2 Hz). This disturbance will seriously affect the cardiac motion extraction. While a small GA 49.8º has shown good performance in terms of both spreading-out the spiral interleaves as well as not affecting the cardiac self-gating extraction.

Cine images acquired with the proposed SMS-SPARCS technique are compared to the clinical standard in Figure 4. LVEF calculations show no significant differences (Figure 5(a)), demonstrating the clinical interchangeability of the two methods. Image quality is slightly lower with the proposed technique (Figure 5(b)), but produced diagnostic quality images.

Conclusion

We have successfully expanded the SPARCS strategy to SMS imaging to further boost scan efficiency. With 10 seconds of acquisition per slice group, whole LV coverage cine images can be obtained around 30 seconds of free breathing without ECG gating.

Acknowledgements

This work is supported by the Fundamental Research Funds for the Central Universities, China, 2022RC20, National Natural Science Foundation of China, 62201081, and National Institutes of Health, United States, R01 H155962. The authors acknowledge the help of our research CMR technologists Xudong Liu.