MRF-gSlider: A High-resolution Multi-parameter Quantitative Magnetic Resonance Imaging Method
Poster No:
2379
Submission Type:
Abstract Submission
Authors:
Dakun Hu1, Huafeng Liu1, Huihui Ye1
Institutions:
1Zhejiang University, Hangzhou, Zhejiang
First Author:
Co-Author(s):
Introduction:
Magnetic resonance fingerprinting(MRF)(Ma et al., 2013) is an emerging technique that can obtain quantitative results of multiple magnetic parameters within a single MR scan. However, the slice thickness of common 2D MRF can only reach 3-5mm(Badve et al., 2017; Liao et al., 2018), which is quite limited for small lesion detection and fine structure characterization. Though 3D MRF(Liao et al., 2017; Ma et al., 2019) enables much higher slice resolution(~1mm), it suffers reconstruction burden from large data of 3D non-cartesian trajectory acquisition.
We introduced the idea of gSlider(Generalized SLIce Dithered Enhanced Resolution)(Setsompop et al., 2018), which has a remarkable resolution improvement effect in diffusion MRI, into MRF, and proposed MRF-gSlider, to achieve high-resolution multi-parameter quantitative magnetic resonance imaging.
We introduced the idea of gSlider(Generalized SLIce Dithered Enhanced Resolution)(Setsompop et al., 2018), which has a remarkable resolution improvement effect in diffusion MRI, into MRF, and proposed MRF-gSlider, to achieve high-resolution multi-parameter quantitative magnetic resonance imaging.
Methods:
Based on the idea of gSlider, as shown in Fig.1, we designed 4 RF pulses and regarded the slice profile as composed of 4 sub-slices, where for each excitation, one sub-slice undergoes a π phase modulation. Gslider signals obtained after scanning with these four RF pulses are essentially different weighted combination of four sub-slice signals, which can be expressed in the form of a matrix as
Y4×N×t=A4×4S4×N×t (1)
Among them, Y represents the 4 gSlider signals, S represents the independent signals of four sub-slices, A represents the weighted coefficient matrix, N represents the number of pixels in the image, and t represents the number of time points of the signal. Given A, we can easily solve for the independent signals of each sub-slice. Then perform dictionary matching separately, we can obtain quantitative results with 4 times higher resolution than common MRF. And A can be calculated through simulation experiments. We formulated 4 gSlider signals and 4 sub-slice signals from the dictionary based on the excitation profiles shown in Fig.1 and preset T1 and T2 maps. Then put the simulation signals into (1), we obtained the A matrix.
To evaluate our method, we conducted multiple sets of scans. In phantom scans, We mainly validate the quantitative accuracy and uniformity across the slices of the T1 and T2 maps. Moreover, we also validate the robustness and repeatability. In vivo scans performed on a healthy adult's brain. Here we concerned about whether subtle structural changes in slice direction can be clearly distinguished.
All scans were performed on a Siemens 3T Prisma scanner with a 20-channel coil based on a 2D FISP MRF sequence with gSlider pulses. Spiral trajectories and 36x downsampling were used to speed up the acquisition process with the in-plane resolution at 1x1mm². A single acquisition takes about 20 seconds. The sub-slice thickness achieved is 1.5mm.
Y4×N×t=A4×4S4×N×t (1)
Among them, Y represents the 4 gSlider signals, S represents the independent signals of four sub-slices, A represents the weighted coefficient matrix, N represents the number of pixels in the image, and t represents the number of time points of the signal. Given A, we can easily solve for the independent signals of each sub-slice. Then perform dictionary matching separately, we can obtain quantitative results with 4 times higher resolution than common MRF. And A can be calculated through simulation experiments. We formulated 4 gSlider signals and 4 sub-slice signals from the dictionary based on the excitation profiles shown in Fig.1 and preset T1 and T2 maps. Then put the simulation signals into (1), we obtained the A matrix.
To evaluate our method, we conducted multiple sets of scans. In phantom scans, We mainly validate the quantitative accuracy and uniformity across the slices of the T1 and T2 maps. Moreover, we also validate the robustness and repeatability. In vivo scans performed on a healthy adult's brain. Here we concerned about whether subtle structural changes in slice direction can be clearly distinguished.
All scans were performed on a Siemens 3T Prisma scanner with a 20-channel coil based on a 2D FISP MRF sequence with gSlider pulses. Spiral trajectories and 36x downsampling were used to speed up the acquisition process with the in-plane resolution at 1x1mm². A single acquisition takes about 20 seconds. The sub-slice thickness achieved is 1.5mm.
Results:
Fig.2 shows the quantitative results for the phantom and brain. (a) is a T1 quantitative result of the phantom, including the golden standard and four sub-slice T1 map. Similarly, (b) is the corresponding T2 quantitative result of the phantom. It can be found that the quantitative results of T1 and T2 are highly consistent with the golden standard, and the results among the four sub-slices also show good consistency.
In (c) and (d), The coordinates of each point represent the average quantitative value of a same ROI in two different scans of the phantom. It shows that the results of the two scans are stable and our method is repeatable.
The T1 and T2 quantitative results of the brain are shown in (e) and (f). It can be seen that the subtle structural changes between adjacent sub-slices can be clearly distinguished.
In (c) and (d), The coordinates of each point represent the average quantitative value of a same ROI in two different scans of the phantom. It shows that the results of the two scans are stable and our method is repeatable.
The T1 and T2 quantitative results of the brain are shown in (e) and (f). It can be seen that the subtle structural changes between adjacent sub-slices can be clearly distinguished.
Conclusions:
Based on the idea of gSlider, we proposed a high-resolution multi-parameter quantitative MRI method that combines MRF and slice encoding, which can obtain a 4 times improvement on slice resolution compare with common MRF. Moreover, the proposed method has been validated on both phantom and in vivo scans, show that it can obtain accurate, uniform, and repeatable quantitative results.
Modeling and Analysis Methods:
Methods Development 2
Novel Imaging Acquisition Methods:
Imaging Methods Other 1
Keywords:
MRI PHYSICS
Other - MR Fingerprinting;Quantitative MRI;gSlider
1|2Indicates the priority used for review
Provide references using author date format
Liao, C. et al. (2017) ‘3D MR fingerprinting with accelerated stack-of-spirals and hybrid sliding-window and GRAPPA reconstruction’, NeuroImage, 162, pp. 13–22.
Liao, C. et al. (2018) ‘Detection of Lesions in Mesial Temporal Lobe Epilepsy by Using MR Fingerprinting’, Radiology, 288(3), pp. 804–812.
Ma, D. et al. (2013) ‘Magnetic resonance fingerprinting’, Nature, 495(7440), pp. 187–192.
Ma, D. et al. (2019) ‘Development of high-resolution 3D MR fingerprinting for detection and characterization of epileptic lesions’, Journal of Magnetic Resonance Imaging, 49(5), pp. 1333–1346.
Setsompop, K. et al. (2018) ‘High-resolution in vivo diffusion imaging of the human brain with Generalized SLIce Dithered Enhanced Resolution Simultaneous MultiSlice (gSlider-SMS)’, Magnetic resonance in medicine, 79(1), pp. 141–151.