Cognitive Application of Multi-Phase Passband Balanced SSFP fMRI with 50ms Sampling rate at 7 Tesla
Zhongwei Chen1,2, Rong Xue1, Jing An3, Kaibao Sun1,2, Zhentao Zuo1, Peng Zhang1, and Danny JJ Wang4

1State Key Laboratory of Brain and Cognitive Science, Institute of Biophysics, Chinese Academy of Sciences, Beijing, China, People's Republic of, 2Graduate School, University of Chinese Academy of Sciences, Beijing, China, People's Republic of, 3Siemens Shenzhen Magnetic Resonance Ltd, Shenzhen, China, People's Republic of, 4Laboratory of FMRI Technology (LOFT), Department of Neurology, University of California Los Angeles, Los Angeles, CA, United States


Multi-phase passband steady state free precession (SSFP) cine fMRI can achieve a spatial resolution of a few mm3 and a temporal sampling rate of 50ms at 7 Tesla , while maintaining low geometric distortion and signal dropout. In this study, the feasibility and accuracy of the technique are demonstrated by two visual event-related functional MRI experiments.


We present a new cognitive application of a self-developed multi-phase segmented passband balanced steady state free precession (b-SSFP) cine fMRI technique. Compared to standard gradient-echo echo-planar imaging (GE-EPI), one unique advantage of multi-phase segmented b-SSFP fMRI is that it provides an ultra-fast temporal sampling rate on the order of about 10-20Hz. In this study, a self-improved multi-phase passband cine b-SSFP technique was introduced with a sampling rate of 50 ms per volume and a spatial resolution of a few mm3, which can be applied to a visual event-related functional MRI (ER-fMRI).(1,2)


ER-fMRI experiments were performed on a 7 Tesla Siemens Magnetom whole-body system (Erlangen, Germany) with a volume excitation 24-channel receive Nova Medical head coil. ER-fMRI was performed with a 16 Hz-flashing counter-phased checkerboard (duration 62.5ms) as tvisual stimuli. In Exp.1, a TTL trigger was sent to the stimulus computer at the beginning of each trial and the checkerboard was presented with a time delay of 0, 100, 200 or 300ms (d0, d100, d200, d300 respectively) relative to the TTL trigger pulse (Fig.1). In Exp. 2, visual stimuli were presented in the left and right visual field sequentially in each trial, one immediately at the TTL trigger and the other on the opposite visual field with a delay of 100ms, 200ms, or 300ms (d100, d200, d300 respectively) relative to d0 (Fig.1). Each experiment was carried out on 9 subjects.


A general linear model (GLM) was employed to detect the spatial localizations of the initial dip and subsequent positive hemodynamic response function (HRF). The activation maps of the positive and negative HRF activations from a representative subject are plotted in Fig.2a (Family Wise Error (FWE) corrected t-test, p<0.05). A significant initial dip was detected in each subject. The averaged and smoothed HRF curve (with standard deviation) is displayed as percentage signal changes of the subsequent time series of b-SSFP images for the stimuli without delay relative to the TTL trigger (d0 condition), as representative, shown in Fig.2b. In Exp.1, the HRF curves from visual stimulation with four different delays (d0, d100, d200, d300) relative to the TTL trigger pulse are displayed in Fig. 3a respectively. The BOLD response functions of Exp.2 are shown in Fig.3b, in which the HRFs of 4 stimulus conditions (d0, d100, d200, d300) are presented. Each condition contained two semi-screen stimuli sequentially presented in the left and right fields respectively with one of the 4 temporal delays between them. Fig.3c shows the average estimated fMRI delay with error bars of standard deviation between HRFs of each condition in Exp.1 and 2. Repeated measures ANOVA of estimated HRF delays of fMRI revealed a significant statistical difference across four delay conditions in Exp.1 (p=0.00006) and Exp.2 (p=0.00009), and the post-hoc test showed that there were significant differences between each pair of two conditions (p<0.00001). Repeated measures ANOVA also showed significant difference in HRF delays between Exp.1 and Exp.2 (p<0.0001).


In this study, a new fMRI technique was presented by combining passband b-SSFP with multiphase cine acquisitions to achieve a temporal sampling rate of a few tens milliseconds and a spatial resolution of a few cubic millimeters. The technique inherits the advantages of passband b-SSFP, such as high SNR efficiency and relatively low magnetic susceptibility artifacts, and further capitalizes on the unique advantage of b-SSFP for steady state imaging to allow for a high temporal sampling rate. The resultant HRF curve provides reliable estimate of the initial dip, and is able to discern a stimulus delay as short as 100ms. In other words, the multiphase b-SSFP technique is able to differentiate 100ms, 200ms and 300ms delays of visual stimulus presentation.


In conclusion, multi-phase b-SSFP offers a promising approach for high fidelity fMRI with millisecond temporal sampling rate and millimeter spatial resolution, especially at ultrahigh magnetic fields. With accelerated imaging speed, the technique may find a range of applications in cognitive and clinical neuroscience in the future.


The work is supported in part by the Ministry of Science and Technology of China (MOST) grants (2012CB825500), National Nature Science Foundation of China grant (91132302), Chinese Academy of Sciences Strategic Priority Research Program B grants (XDB02010001, XDB02050001), and US NIH grants (R01-EB014922, R01-NS081077).


(1)Chen Z, et al. ISMRM. 2014; p4202. ( 2) K L.Miller. FMRI using balanced steady-state free precession (SSFP). Neuroimaging. 2011;62:713-719.


Fig.1 Event-related fMRI designs of Exp.1 and Exp.2.

Fig.2 (a) Positive activations of SPM-t map onto segmented Mprage-T1 structure image in d0 condition (as an example) of Exp.1 (Left). The initial dip in SPM-t map onto the same segmented structure image (right). (b) The averaged and smoothed HRF curve, averaged by 9 subjects, with error bar (standard deviation) is plotted.

Fig.3 The BOLD response functions averaged by 9 subjects in Exp.1 (a) and Exp.2 (b) are shown. Four conditions (d0, d100, d200, d300) of delay are presented, (c) The figure shows the averaged estimated fMRI delay with error bar (standard deviation) between HRFs of every condition.

Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)