Improved tractography of post mortem human brain at 7T using DWSSFP

Sean Foxley, Saad Jbabdi, Wilfred Lam, Olaf Ansorge, Gwenaelle Douaud, and Karla Miller FMRIB Centre, University of Oxford, Oxford, OXON, United Kingdom, University of Oxford, Oxford, OXON, United Kingdom INTRODUCTION: Post-mortem human brain imaging is of increasing interest for in-vivo validation studies. Lack of time restraints enables acquisitions with very high spatial resolution, leading to improved identification of white matter tracts.1 Use of diffusion weighted steady-state free precession (DW-SSFP)2 has been demonstrated to perform significantly better than diffusion-weighted spin echo techniques3 for post-mortem human brain tractography at 3T. However, even with long scan time, the reductions in T2 and diffusion coefficients in fixed tissue conspire to make secondary fibre populations very difficult to estimate. In this work we evaluate several improvements to our previous imaging protocols: (1) increased field strength and increased receive channels, both of which boost signal levels; and (2) higher b-value, which boosts diffusion contrast. METHODS: A cohort of n=4 fixed, post-mortem human brains were scanned to compare our previously optimized protocol at 3T with a newly optimized 7T protocol (table 1). For all protocols, DW-SSFP was acquired at 1mm isotropic resolution with diffusion weighting along 50 directions (repeated twice for total scan time 18-24 hours). While b-value is not defined in DW-SSFP, we can set the protocol q-value to achieve a desired level of contrast based on tissue T1, T2, and ADC. Hence, we targeted q=300 cm-1 to achieve the effective contrast of b=4000 s/mm2. Following an established pre-processing pipeline1, first and second fibre orientation populations were estimated using a modified version of BEDPOST that includes the DW-SSFP signal model. Tractography was performed with PROBTRACKX2. All tractography was performed with identical seed masks between the 3T and 7T data. RESULTS: Angular uncertainty in the primary diffusion direction (PDD) is show for both 3T and 7T data in figure 1. The peak of the uncertainty in the 7T (figure 1 inset) is lower and with a higher proportion of voxels than 3T, indicating that 7T data more accurately estimate the PDD. Figures 2-5 demonstrate typical maximum intensity projections (MIP’s) of tractography results. For all figures, column 1 is 3T data and column 2 is 7T. Figure 2 shows the genu of the corpus callosum (CC) projected down the axial direction, where 3T and 7T data are quite similar. This portion of the CC has few radiating tracts in to the cortex that cross the centrum semiovale (CS). Figures 3-5 demonstrate three major tracts that interdigitate through the CS: the corticospinal tract (CST), regions III and IV of the CC, and the superior longitudinal fasciculus (SLF). Without sufficient crossing fibre information, these tracts will often terminate at this crossing point or mistrack along the primary diffusion direction. In all cases the 7T results show more cortical projections, which is indicative of improved secondary fibre estimates4 . 3T data all suffer from few seconday fibres across the CS and consequently do not have the characteristic fanning anatomy seen in the 7T data. Specifically, figure 3 shows the CST (note this includes other motor related projection tracts). MIP’s were produced down the xand yaxes, demonstrating the drastic improvement in general tracking as well as the increase in cortical projections. Figure 4 shows regions III and IV of the CC. Much like the CST, 7T data provide much better tracking across the CS than the 3T counterpart. In fact, 3T results do not even completely reach the superior cortex, CC III and IV’s primary terminal region. Finally, figure 5 demonstrates the projection of the left SLF. As with the previous tracts, the 7T data demonstrates a more complete tract with more projections than the 3T results. DISCUSSION: These results provide evidence for the improvement of DW-SSFP at 7T compared to 3T in post-mortem brains. Higher fidelity of tractography demonstrates an overall improvement in data quality and diffusion-dependent measurements. Since tractography is improving with each iteration of methodological development, post-mortem diffusion weighted imaging with SSFP is increasingly presenting itself as an ideal method for probing more complicated and interdigitated white matter tracts. References: [1] Miller, NeuroImage, 2011, [2] Buxton, MRM, 1993, [3] Miller, NeuroImage, 2012, [4] Foxley, ISMRM, 2012 Protocol δ/TE/TR Flip Gdiff beff 3T q300 19/26/35 ms 35 ̊ 3.8 G/cm 4000 s/mm2 7T q300 13/21/30 ms 30 ̊ 5.6 G/cm 4000 s/mm2 Table 1 Figure 1 Figure 2