A DOUBLE ACQUISITION FOUR-CONTRAST IMAGING APPROACH TO DELINEATE HUMAN BRAINSTEM ANATOMY IN-VIVO AT 7 TESLA
Michael Wyss1, Laetitia Vionnet1, Mike Bruegger1,2, Bernd Daeubler3, Lars Kasper1,4, Daniel Nanz5, Marco Piccirelli3, David O. Brunner1, and Klaas P. Pruessmann1

1Institute for Biomedical Engineering, University of Zurich and ETH Zurich, Zurich, Switzerland, 2Center of Dental Medicine, University of Zurich, Zurich, Switzerland, 3Department of Neuroradiology, University Hospital Zurich, Zurich, Switzerland, 4Translational Neuromodeling Unit, Institute for Biomedical Engineering, University of Zurich and ETH Zurich, Zurich, Switzerland, 5Institute of Diagnostic and Interventional Radiology, University Hospital Zurich and University of Zurich, Zurich, Switzerland

Synopsis

The human brainstem anatomy is challenging to image in vivo on a single subject basis. The densely-packed interspersion of nuclei and white matter tracts cannot typically be imaged with an image contrast strong enough for differentiation of relevant substructures. We present an MR imaging approach at 7 Tesla that requires two high resolution MR acquisitions. From the two data sets, four image series with varying contrast weightings can be derived and used to delineate brainstem anatomy. The proposed strategy resulted in exceptionally high image quality enabling differentiation of several brainstem substructures that are hardly discernible in commonly acquired MR images.

Purpose

Based on previous own work (1), the main objective of this report was to further improve visualization and delineation of human brainstem substructures in-vivo.

Methods

The study was approved by the local ethics board. Informed consent was obtained from six healthy subjects (mean age: 26, range: 22-29, four female). All experiments were performed on a 7T MRI system (Achieva, Philips Healthcare, Cleveland, OH, USA) using a 32 channel receive coil (Nova Medical, Wilmington, DE). An inversion recovery 3D MPRAGE sequence was acquired: FOV: 230 x 230 mm2, TR for Inversion: 5 s, TR: 7.9 ms, TE: 3.6 ms, flip angle: 5°, voxel size: (600μm)3, 90 slices, 1 signal average, SENSE factor 2.0, Bandwidth: 202 Hz/pixel, acquisition time: 09:05 minutes. The acquisition delay was set to 970 ms for gray matter nulling. This scan was repeated four times with the same settings which resulted in a total scan time of approximately 36:20 minutes. To account for displacement of anatomy, intra-subject rigid body realignment over the repetitions was performed. Subsequently the average of the four aligned datasets was calculated. Second, a 3D T2*-weighted multi-gradient-echo (6 echoes) sequence was acquired: FOV: 230 x 230 mm2, TR: 60 ms, first TE: 5.5 ms, delta TE: 6.0, flip angle: 18°, voxel size: (600μm)3, 90 slices, 1 signal average, SENSE factor 2, Bandwidth: 518 Hz/pixel, acquisition time: 16:26 minutes. The multi-echo MR signal intensity as a function of the echo time, $$$S(TE)$$$, was fitted pixel wise to the expression $$$lnS(TE)=lnS(TE=0)-R_2^**TE$$$ in a linear least-squares polynomial regression. T2* was obtained by inversion: $$$T_2^*=\frac{1}{R_2^*}$$$. Finally the first four echo images of the T2* weighted sequences were combined to a single image according to the concept of multiple-echo data image combination (MEDIC). The images were visually compared to histology and post-mortem MR images from the Duvernoy's Atlas (2) and Olzsewski and Baxter's Cytoarchitecture of the Human Brainstem (3).

Results

Displacements during the MPRAGE acquisitions were minute for in plane movement with maximal shifts of 0.2 mm (X-axis), 0.3 mm (Y-axis) but substantial in the Z-axis with maximal shifts of 1.8 mm. Rotation of the head was below one degree in the jaw angle with mean values of max. rotation of 0.48° ± 0.36°. The rotational changes in the pitch and roll angle were negligible (Fig. 1). Figure 2 shows the reformatted sections in three orientations of the isotropic MPRAGE dataset of one subject. Figure 4 illustrates all four contrasts of one subject at three levels of the brainstem (midbrain, pons, medulla oblongata) without annotation. Brainstem substructures were manually outlined by comparing to histology sections from the Duvernoy Atlas, as illustrated in figure 3 and 5. The identifiable structures were assigned as follows: 1substantia nigra 2olivary nucleus, 3superior colliculus, 4inferior colliculus, 5medial lemniscus, 6mammillary body, 7red nucleus, 8decussation of the superior cerebellar peduncle, 9medial longitudinal fasciculus, 10mamillothalamic tract, 11posterior commissure, 12periaequaductul gray matter, 13lateral lemnisucs, 14optic tract, 15superior cerebellar peduncle, 16ventral pontine decussation, 17inferior and lateral vestibular nuclei / spinal trigeminal nuclei and tract, 18pyramidal tract, 19medial longitudinal fasciculus / tectospinal tract, 20inferior cerebellar peduncle, 21thalamic fasciculus, 22lenticular fasciculus, 23corticospinal tract, 24spinal trigeminal nucleus and tract, 25raphe nuclei, 26pontine fibers.

Discussion

The isotropic MPRAGE sequence in the gray matter nulling regime provides an ideal contrast to identify small substructures within pons and medulla regions. T2* and R2* maps, not affected by inhomogeneity of the transmit field (B1+) and varying receive-coil sensitivity profiles provide a favorable contrast for structures within the midbrain area. Image combination based on MEDIC resulted in improved SNR and enhanced T2* weighting. This facilitates the visualization of certain brainstem substructures hardly visible on the basis of MPRAGE/ T2*/R2* maps.

Conclusion

Combining a segmented high resolution isotropic 3D MPRAGE in the gray matter nulling regime with T2*/R2* maps and MEDIC calculation resulted in improved visualization and differentiation of the complex human brainstem architecture in-vivo. Further effort is needed to adapt such an approach for standard clinical applications, particular including an optimized measurement time.

Acknowledgements

References

1. Wyss M, Bruegger M, Daeubler B, Vionnet ML, Brunner DO, Pruessmann KP. Visualization of human brainstem substructures using gray matter nulling 3D-MPRAGE at 7Tesla. In: Proc. Intl. Soc. Mag. Reson. Med. 22 (2014). Vol. 4632. Milano, Italy; 2014.

2. Naidich TP, Duvernoy HM, Delman BN, Sorensen AG, Kollias SS, Haacke EM. Duvernoy’s Atlas of the Human Brain Stem and Cerebellum. Vienna: Springer Vienna; 2009. http://link.springer.com/10.1007/978-3-211-73971-6

3. Büttner-Ennever JA, Horn AKE, Olszewski J, editors. Olszewski and Baxter’s cytoarchitecture of the human brainstem. 3rd, revised and extended edition. Basel: Karger; 2014. 290 p.

Figures

Figure 1:
Illustrated are motion estimates (calculated in SPM) before rigid body realignment for the four MPRAGE acquisitions.


Figure 2:

Selected reformatted sections from the isotropic MPRAGE dataset in the sagittal (a), transversal (b) and coronal (c) plane, without annotations. Selection criteria were the optimal presentation of relevant and identifiable brainstem substructures.


Figure 3:
Substructures identifiable on MPRAGE sections.


Figure 4:
All four contrast images (MPRAGE, R2*, T2*, MEDIC) obtained from a single subject at three levels; midbrain (a), pons (b) and medulla oblongata (c) without annotations.



Figure 5:
For the midbrain, R2*, T2* and MEDIC images demonstrate better visibility for 7red nucleus, and 1substantia nigra. 12Periaequaductal gray matter, 26raphe nucleus, 13lateral lemniscus and 15optic tract are best visible on the MPRAGE contrast. Within the pons the 9medial longitudinal fasciculus is well depictable on MPRAGE and MEDIC. Best visibility of 15superior cerebellar peduncle and 23pontine fibers is achieved on MEDIC images. 2Olivary nucleus appears optimal on MPRAGE and MEDIC.




Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
3309