Anna Crawford1, Stephen Jones1, and Mark Lowe1
1Imaging Institute, Cleveland Clinic Foundation, Cleveland, OH, United States
Synopsis
Many advanced imaging application
require the identification of seed and target ROIs. Due to size, and signal
characteristics some brain structures including brainstem nuclei are difficult
to identify with MR imaging, while other types of imaging such as PET have
other drawbacks such as poor spatial resolution. We have come up with a “recipe”
for defining the ventral tegmental area (VTA) as well as the dorsal raphe
nucleus (DRN) using three anatomic atlases. The definition of these ROIs were
confirmed with PET imaging. Our method allows for robust, reproducible ROIs
using only a high resolution T1 weighted image.
Introduction
Many cutting edge applications of advanced imaging, such as
resting state connectivity and HARDI connectivity, necessarily require
identification of seed and target ROIs. While larger brain structures, such as
the sensorimotor cortex can be reliably identified in most humans using MRI,
smaller structures such as dopaminergic and serotonergic nuclei are more
difficult to identify, both due to their small size, and signal characteristics
that are similar to adjacent brain tissue. Another alternative is using nuclear
medicine tracers to identify these small foci, but the drawback is relatively
poorer spatial resolution of PET systems, and the subsequent need to
co-register the two modalities. Therefore, there is great utility is providing
well-defined “recipes” on how to locate these small nuclei in conventional
MRI. We present here two such methods
for identification of the ventral tegmental area (VTA) and dorsal raphe nucleus
(DRN) of the brainstem.Methods
A gold standard for
locating the VTA and DRN was based on three anatomic atlases: Olszewski and
Baxter's Cytoartchitecture of the Human Brainstem 2014, Paxinos and Huang's
Atlas of the Human Brainstem 1995, and Duvernoy's Atlas of the Human Brain Stem
and Cerebellum 2009. Both nuclei were drawn for each atlas. Using a grid of
1mmx1mm for each available slice, the nuclei were filled in if at least half of
the grid block was full. Then the 3 atlases were merged together to form the
DRN and VTA. This merged version was then 'smoothed' such that the nuclei as a
whole would have a cohesive shape and removing any extraneous pieces not in a
cluster.
For both nuclei, the brain volume is oriented in the
plane that runs through the middle of the mammillary body and the superior
colliculus. If the head is rotated left or right, this should also be
corrected.
For the DRN, all parts, ie ventral, dorsal, and
intrafasicular parts were included. The DRN is 8mm thick. The inferior most
level is at the isthmus, where the cerebral aqueduct forms into the 4th
ventricle. The DRN lies at the midline of the brainstem and should follow along
the cerebral aqueduct with the cerebral aqueduct being its posterior border.
See Figure 1 for diagram.
For the VTA, the paranigral, interfasicaular, central
linear, parabrachial pigmented, and rostral linear nuclei were included. These
nuclei were included based on their projections to the limbic area, thalamus,
and forebrain as well as the midline nuclei. The VTA is 6mm thick and extends
inferiorly 2mm below the inferior red nucleus, but does not cross the
decussation. The superior extent is approximately aligned with the inferior
mammillary body. The VTA also lies at the midline of the brainstem with the interpeduncular
fossa as its anterior border. See Figure 2 for diagram.
A validation of the
above recipe to localize the DRN and VTA was performed with positron emission
tomography studies that were made available from a previous nuclear medicine
study. [18F]-MPPF scans were used to localize DRN in MRI scans
co-localized to the PET imaging data. [11C]-FeCIT scans, used to
measure dopamine transporter activity, were used to localize VTA in MRI scan
similarly co-localized. ROI’s for each VTA and DRN were determined using the
above described methods using the co-localized MRI scans and compared for
spatial overlap.Results
Figures 1
and 2 show the anatomical delineation of the ROI’s using the prescribed method.
Figures 3 and 4 show a clear overlap of DRN and VTA ROI’s defined using the
landmark method described here and those determined using the relevant PET
ligand studies.Discussion
We have
described a straightforward, anatomic landmark based methodology to produce
ROI’s in the VTA and DRN of the brainstem. The method requires only a high
resolution T1 weighted scan to produce the ROI’s. The landmarks employed are
all conspicuous objects readily identifiable on MRI and the method is simply
described to enable even researchers that are relatively naïve to cross
sectional anatomy to produce robust, reproducible ROI’s.Conclusion
Based on
multiple brainstem atlases, we provide “recipes” to locate the DRN and VTA on
MRI anatomical imaging. These steps are useful for connectivity analysis that
require these small nuclei as endpoints.Acknowledgements
This work was supported
by Siemens Healthineers, Inc. The authors gratefully acknowledge the help of
Tobias Kober of Siemens Healthineers for use of WIP944 (MP2RAGE).
References
-
Olszewski
and Baxter. Olszewski and Baxter's
Cytoartchitecture of the Human Brainstem.Basel, Switzerland: Karger,
2014. pp. 18-19, 44-56, 203-206, 217-218.
- G.
Paxinos and XF Huang. Atlas of the
Human Brainstem. 1995 ed. San Diego, California: Academic Press, 1995.
Fig 50-62.
- T.P. Naidich, H.M Duvernoy, et
al. Duvernoy's Atlas of the Human
Brain Stem and Cerebellum. Austria: Springer-Verlag/Wien, 2009. pp.
55, 84-89