Synthetic MRI with water suppression technique  to reduce CSF partial-volume artifacts
Tokunori Kimura1, Yuki Takai2, Hiroshi Kusahara2, Hitoshi Kanazawa2, and Ryo Shiroishi3

1Department of Radiological Sciences, Shizuoka College of Medicalcare Science, Hamamatsu, Japan, 2MRI development department, Canon Medical Systems corp., Otawara, Japan, 3Clinical Research and Development Center, Canon Medical Systems corp., Otawara, Japan


We proposed a new synthetic-MRI technique combined with water suppression to reduce CSF partial volume effects (PVE) artifacts problematic in a conventional synthetic-MRI. Our water suppression was simply achieved by subtracting additionally acquired long-TE SE image of water signal dominant. After the quantitative parameter maps of original and with water suppression were generated, water-suppressed synthetic-SE and -FLAIR images were calculated using those suitable combinations. We demonstrated that CSF PVE artifacts were dramatically reduced in our proposed synthetic-FLAIR, and furthermore that, by the two-compartment model simulation and volunteer MR brain study, our synthetic-SE provided better gray-white matter contrasts compared to our synthetic-FLAIR.


A synthetic MRI (SyMRI), that generates several kinds of contrast-weighted images by synthesizing quantitative parameter maps of proton density (PD), T1-longitudinal relaxation time (T1), and transverse relaxation time (T2) [1] obtained using acquired fast-spin echo (FSE) data, is beginning to be used clinically because of the advantages of saving total examination time and the freedom to select MRI acquisition parameter of repetition time (TR), inversion time (TI), and echo time (TE). However, a synthetic T2-weighted (T2W) Fluid-attenuation inversion recovery (FLAIR) introduces hyper intense artifacts at the border zone of tissue and ventricle or the surface of the brain, and it is considered that the cause is partial volume effects (PVE) of brain tissue and cerebral spinal fluid (CSF) [2,3]. The purpose of this study was to propose and assess a new synthetic SE and FLAIR technique combining with water suppression technique to reduce those artifacts.



When a unit voxel consists of two components of water and tissue, MR signal in the voxel is based on a two-compartment model as shown in Fig.1. Our proposed water suppression was based on the technique of subtracting additionally acquired long TE data of water signal dominant from the shorter TE data [4]. At least 3-echo data are acquired with the same TR and/or TI. A processing flow for our proposed technique of water-suppressed synthetic MRI using minimum number of data is shown in Fig. 2.


The parameters in this simulation were summarized in Table.1. Voxel mean values of quantitative parameters of PD, T1, and T2 were calculated from those averaged signal intensities as a parameter of a water volume fraction, Vw and a subtracting weight, ‘α’, then synthetic-SE and -FLAIR signal intensities were calculated as a function of TE each with a parameter of Vw, where the original T1 was commonly used since the FLAIR can suppress water Mz even using original T1.

MRI Experiments

In MR experiments, the first four data in Fig. 2 were acquired for our proposed SyMRI. A healthy volunteer study was performed on a 3T MRI (‘Galan 3T ZGO’, Canon Medical Systems corp., Otawara, Japan) with a 32-channel head coil after obtaining informed consent. A fast spin echo sequence was used and the acquisition parameters were: parallel imaging (SPEEDER) of speed-up factor 2, acquisition matrix of 256x256, display matrix of 512x512 after sinc interpolation, FOV=23cm, slice thickness=5mm, the number of slices selected at maximum, NAQ=1, TR1=4000ms, TE1=20ms, TE2=100ms, TE3=300ms, TI1=1000ms, and an adiabatic inversion pulse for IR to reduce B1 inhomogeneity.



The quantitative parameters for the water-suppressed data, compared to the original (α=1) data, became close to those for tissue components, and the water suppression effects were stronger with increasing subtracting weight, ‘α’ (Table 1). For the signal intensities as a function of TE (Fig.3), the cause of hyper intense artifacts due to PVE in conventional synthetic-FLAIR was clarified and those artifacts were reduced in our proposed water-suppressed synthetic-FLAIR. Furthermore, our proposed synthetic-SE provided SNR improvements in addition to CSF suppression.

MR experiments

As shown in Figure 4, The signal intensities in CSF portions were reduced with increasing the subtracting weight, α, both for the synthetic-SE and -FLAIR. For synthetic-FLAIR of original (α=1), the border area between tissue and ventricles and the narrow portions such as the surface of brain were dominant; in contrast, those hyperintense artifacts were not obvious in our proposed water-suppressed images even at longer TE. Furthermore, the gray-white matter contrast was better on the synthetic-SE than on the synthetic-FLAIR and the acquired FLAIR.


We proposed a new synthetic-SE and -FLAIR techniques based on a simple water suppression using additional long TE data to reduce CSF PVE artifacts in conventional synthetic MRI. We demonstrated that the artifacts were dramatically reduced by simulation and volunteer MR brain study. Alternative approaches to reduce CSF PVE artifacts in SyMRI were proposed such as a 3D acquisition [5] or a combination with deep learning (DL) technique [6]. However, the PVE artifacts cannot be principally avoided when using small voxel, and the DL is complicated procedure and the robustness of suppression is unclear. In contrast, our proposed technique is very simple and easily combined to the conventional SyMRI technique.


Although further optimization of TE for water image or processing techniques to optimally suppress CSF signals while minimizing SNR reduction due to subtraction, our proposed water suppression technique will solve the problem of CSF PVE artifacts in current synthetic-FLAIR or -SE, and is expected to become clinically further useful.


No acknowledgement found.


[1] Warntjes JB, et al. Rapid magnetic resonance quantification on the brain: optimization for clinical usage. Magn Reson Med. 2008;60:320–329.

[2] Hagiwara A et al. SyMRI of the Brain: Rapid Quantification of Relaxation Rates and Proton Density, With Synthetic MRI, Automatic Brain Segmentation, and Myelin Measurement. Investigative Radiology. 2017; 52:647–657.

[3] Tanenbaum LN et al. Synthetic MRI for Clinical Neuroimaging: Results of the Magnetic Resonance Image Compilation (MAGiC) Prospective, Multicenter, Multireader Trial. AJNR 2017;38:1103-1110.

[4] Essig M et al. Assessment of cerebral gliomas by a new dark fluid sequence, high intensity REduction (HIRE): a preliminary study. JMRI. 2000;11:506-517.

[5] Hwang KP et al. 3D isotropic multi-parameter mapping and synthetic imaging of the brain with 3D-QALAS: Comparison with 2D MAGIC. ISMRM2018 #5627.

[6] Gong E et al. Improved Synthetic MRI from multi-echo MRI Using Deep Learning. ISMRM2018 #2795.


Fig. 1. Single- and Two-compartment MR signal models containing water and tissue components in a single voxel.

Fig. 2. Calculation flow for our proposed water-suppression Synthetic MRI The data #5 is required only for T1wsup but not used to make our synthetic images.

Table 1. Simulation parameters (left) and the results of the quantitative parameters in the original and water suppression (wsup) as a parameter of a water volume fraction,'Vw' and the subtracting weight, 'α' (right). Note that the quantitative parameters with ‘wsup’ became close to those of tissue with increasing α.

Fig. 3. Simulation results of signal intensities vs. TE as a parameter of water volume fraction (Vw) Ideally acquired FLAIR(a), conventional synthetic-FLAIR (b), our proposed synthetic-FLAIR (c), andour proposed synthetic-SE (d) both with water suppression (α=1). Note that the synthetic -FLAIR signals using original parameters were overestimated at Vw of 25~75% with increasing TE (b). In contrast, the synthetic-FLAIR signals with (PDorig, T1orig, T2wsup) (c) were close to the ideal FLAIR signals (a). The synthetic-SE signals with (PDwsup, T1orig, T2wsup) (d) were close to a and c except for Vw=100%; furthermore, the tissue SNR became better than those.

Fig. 4. Volunteer results of acquired images (a) (blue frame was used), quantitative parameter maps of PD and T2 and synthetic images shown as a parameter of the subtraction weighting, α (b), and TE dependency of the synthetic-FLAIR (c). Note that the values in PD- and T2-maps and the signal intensities in CSF portions were dramatically reduced with increasing ‘α’. For synthetic-FLAIR, the border area between tissue and ventricles (arrows) and the surface of brain (circle) were dominant on the original with increasing TE; in contrast for the water suppression, those artifacts were reduced with increasing α and negligible even at long TE.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)