T2 mapping of ventricular cerebrospinal fluid (CSF) at 7 Tesla with comparison to 3T
Jolanda M Spijkerman1, Esben T Petersen2, Peter R Luijten1, Jeroen Hendrikse1, and Jaco J M Zwanenburg1

1Radiology, University Medical Center Utrecht, Utrecht, Netherlands, 2Centre for Functional and Diagnostic Imaging and Research, Danish Research Centre for Magnetic Resonance, Copenhagen University Hospital, Hvidovre, Denmark


The relaxation parameters of CSF may have potential as imaging markers in several diseases. In this work T2 mapping of CSF in the lateral ventricles and in the fourth ventricle was performed in six volunteers at 7T, with comparison to 3T. The sensitivity for B1 was assessed by comparing the T2s in both regions, with equal B1 at 3T and different B1 at 7T. T2 values were significantly lower at 7T compared to 3T. No significant difference was found between the lateral and the fourth ventricular T2s at 7T (and 3T), indicating negligible B1-sensitivity for the used T2-mapping sequences.


The relaxation parameters T1 and T2 of cerebrospinal fluid (CSF) depend on its content and on the MRI field strength [1] and may, thus, have potential as an imaging marker in several diseases. The T1 of CSF at 7T is known [2]. The T2 of CSF has been reported as part of a relaxation parameter mapping study at 7T, but the same study considered the reported CSF parameters not reliable because the used method was not robust against the pulsatile flow of CSF [3]. In this work, a flow insensitive T2 mapping method was used to measure the T2 of CSF at 7T and compare it to 3T. The sensitivity to B1 was assessed by comparing the T2 in two regions with different B1 at 7T, relative to the T2 in the same regions at 3T (where the B1 was not significantly different).


An MLEV-based pulse sequence for T2 mapping of CSF was used as described by Qin (2011) [4], using non-selective preparation pulses, which is intrinsically flow-insensitive. Effective echo times of 600, 1200, 2400, 4800ms were used in all experiments. Additionally, a plain, single echo SE-EPI sequence was used for comparison, using the following echo times: 240, 285, 420, 645, 960, 1365, 1860, 2445, 3120, 3885, 4740ms. In all scans two single slice scans were acquired with 1x1x4 mm3 and 3x3x6mm3 resolution, SENSE 2.3, TR 15s, FOV 240x240x4mm3, using a single shot 2D SE-EPI readout with TE 344ms and 53ms for the higher and lower resolution scans, respectively. For all volunteers B1 maps were acquired [5], with 1x1x4mm3 and 3x3x6mm3 resolution.
Six volunteers (aged 21-45 years, 3 male) were scanned with both T2 mapping sequences, at 3T (Philips Healthcare) with an 8ch head coil and at 7T (Philips Healthcare) with a 32ch head coil (Nova Medical). The scans planning is shown in Figure 1A. Conservative region-of-interest (ROI) masks of the lateral ventricles and fourth ventricle were obtained, avoiding partial volume effects with tissue (Figure 1BC). Peripheral CSF was not studied, as these values suffer from partial volume effects with tissue (data not shown).
In order to minimize the effect of noise, signal intensities within the ROIs were averaged before estimating the T2 values from a single exponential decay model. Also mean B1 in the ROIs was determined. Paired t-tests were used to compare T2- and B1-values.


T2 values and B1 values are summarized in Table 1. For both sequences and for both resolutions the measured T2s are significantly shorter at 7T compared to 3T. At both field strengths and for both resolutions, the SE-EPI sequence results in a shorter T2 than the MLEV sequence. No significant differences were found between the T2 values of the lateral ventricles and the fourth ventricle at 7T or 3T. At 3T, the B1 values in the lateral ventricles and the fourth ventricle were similar, but at 7T the B1 was significantly lower (by approximately 20%) in the fourth ventricle compared to the lateral ventricles. For the MLEV-based sequence the T2 values of the high- and low-resolution scans were similar, but for the SE-EPI scans significant differences were found between the resolutions.


The measured T2 of CSF is significantly lower at 7T compared to 3T, which may be caused by a field strength dependent relaxivity of the content (proteins and/or O2) of CSF [6]. These results seem to show a stronger field strength dependency for T2 than for T1 [2], though a side-by-side comparison is still needed.
The SE-EPI sequence resulted in shorter T2s than the MLEV-based sequence, which is probably due to diffusion and flow effects: the MLEV-based sequence is less diffusion- and flow sensitive due to the use of multiple, non-selective refocusing pulses. Since similar T2 values were found for the high- and low-resolution MLEV scans, despite the longer readout TE in the higher resolution scan, this method seems indeed robust for flow effects. With the current parameters the SE-EPI sequence is more vulnerable for partial volume effects compared to the MLEV sequence (data not shown), which likely explains the significantly lower T2 values for the low resolution SE-EPI scans.
No significantly different T2 values were found in the lateral ventricles and the fourth ventricle, at both field strengths. Since the B1 at 7T was different for the fourth ventricle, these results show that both T2-mapping sequences are relatively insensitive to B1 variation.


The measured T2 values of CSF were significantly lower at 7T compared to 3T, suggesting a field strength dependent relaxivity of the content of CSF.


This work was supported by the European Research Council, ERC grant agreement n°337333.


1. Hopkins, at al., Multiple field strength in vivo T1 and T2 for cerebrospinal fluid protons. Magn Reson Med, 1986. 3: p. 303-311.

2. Rooney, et al., Magnetic field and tissue dependencies of human brain longitudinal 1H2O relaxation in vivo. Magn Reson Med, 2007. 57(2): p. 308-18.

3. Wyss, et al., Relaxation parameter mapping adapted for 7T and validation against pptimized single voxel MRS. ISMRM proceedings, 2013. 2464.

4. Qin, A simple approach for three-dimensional mapping of baseline cerebrospinal fluid volume fraction. Magn Reson Med, 2011. 65(2): p. 385-91.

5. Yarnykh, Actual flip-angle imaging in the pulsed steady state: a method for rapid three-dimensional mapping of the transmitted radiofrequency field. Magn Reson Med, 2007. 57(1): p. 192-200.

6. Guo, et al., A Proteomic Analysis of Individual and Gender Variations in Normal Human Urine and Cerebrospinal Fluid Using iTRAQ Quantification. PLoS One, 2015. 10(7).


Figure 1: A: planning of the T2 mapping scans, through the lateral ventricles and the fourth ventricle. B 1-4: T2 MLEV scans for increasing TEMLEV (TEMLEV is 0.6s, 1.2s, 2.4s, 4.8s). C: the used ROIs, the lateral ventricles (LAT, yellow), and the fourth ventricle (FOU, red).

Table 1: Measured T2 and B1 values. a: p<0.05 vs lateral ventricles, b: p<0.05 vs 3T, c: p<0.05 vs SE-EPI, d: p<0.05 vs resolution 1x1x4mm3

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