Full text

,
MRI-Guided Targeting of Deep Brain Circuits:
github.com/IBT-FMI/StereotaXYZ
Horea-Ioan Ioanas1, Bechara John Saab2, Markus Rudin1
1Institute for Biomedical Engineering, ETH and University of Zurich
2Preclinical Laboratory for Translational Research into Affective Disorders, DPPP, Psychiatric Hospital, University of Zurich
animal imaging center
uzh | eth | zurich
,
Background
The increased availability of high-field Magnetic Resonance Imaging (MRI) systems, genetic
vectors, and molecular biological tools enables a more highly resolved study of deep brain
circuits. These circuits - including monoaminergic (e.g. serotonergic, dopaminergic, noradren-
ergic) and neuropeptidergic (e.g. oxytocinergic, vasopressinergic) systems - are prominently
involved in numerous neurological and psychiatric disorders, and are key mediators in swaths
of nonvolitional phenomena in the healthy brain.
The preparatory procedure for experimentally probing the aforementioned systems involves
physical delivery of genetic vectors and/or signal input to deep brain structures. As target
sites become increasingly well resolved, and advanced imaging equipment imposes additional
spatial constraints, na¨ıve atlas-based structure targeting becomes less feasible. We present
an integrated software and bench-workflow solution, which allows offline and animal-specific
MRI-guided targeting of deep brain structures. We document the method’s capabilities on the
basis of data demonstrating a cryogenic-coil compatible slanted targeting of the Dorsal Raphe
(DR) nucleus for optogenetic functional MRI (opto-fMRI).
Workflow
The insertion guiding features of StereotaXYZ can be used agnostic of the animal’s individual
neuroanatomy (in which case StereotaXYZ offers visualization aid and skull variability correction
for canonical atlas-based targeting), or in conjunction with anatomical MR images of individual
animals. MRI sessions are thus optional, but highly recommended if neuroanatomy is expected
to vary in addition to the more generally well documented skull variability [1], which provides
the core rationale for the package.
Session 1 (MRI, optional)
Session 2 (Surgical)
Session 3 (MRI, optional)
Pilot MRI
Mouse Stereotactic
Positioning
Verification MRI
Anatomical Data
StereotaXYZ
Skullsweep
Skullsweep Data
Implatation
Anatomical Data
StereotaXYZ can plot implant guiding figures on such individualized MRI images (figure 6),
on a generic template (figure 5), or on no backdrop at all (figure 4). At its core, StereotaXYZ
computes an implant insertion site and insertion depth for a given angle, target, and skullsweep
file. The only obligatory part of the StereotaXYZ workflow is thus the creation of a skullsweep
file, which is a simple comma separated values file (columns separated by commas, rows by
newline characters) formatted analogously to the example under table 1, and containing the
coordinates of points on the skull. These coordinates can be obtained by lowering a thin needle
or glass pipette until it barely touches the skull.
MRI Set-Up Compatibility
The very low profile implant resulting from the workflow depicted in the right-hand column
handles well in an MRI setting, as show in figures 1 and 2. Conspicuously, the cement cap
extends rostrally far beyond the implant site - this is in order to improve stability, and to relegate
artefacts (most commonly emerging at the edges of the remaining skin tissue) farther from the
cortex.
Figure 1: The angle of the implant places its exterior ferrule precisely in the horizontal plane.
Figure 2: The implant is fully compatible (having minimal thickness and full contact) with rigid non-perforable surface coils (e.g. cryogenic coils).
The image depicts a room temperature coil for better visibility.
Outlook
I”Nearest possible incision” detection.
IMore complex, multi-criteria (e.g. ”avoid region”) path finding.
INonliniar template-to-anatomy transformation.
IImplant performance testing for opto-fMRI.
Session 1
DR Region of Interest
Preoperative Overview
Figure 3: A high-definition MRI scan acquired ahead of the surgical stereotactic procedure can serve to identify the exact neuroanatomical lay-out in the animal; this is particularly relevant for animals with
previous implants, lesions, or when targeting particularly small structures. In the example at hand we plot a generous region of expected distribution of ascending serotonergic cell bodies, of which we chose the
center of gravity as the target in the next session. StereotaXYZ provides registration capabilities ensuring that the image can be co-plotted with skullsweep corrdinates relative to bregma (handled internally via
ANTs [2] and nipype [3], but does not include other preprocessing or conversion tools (for which we recommend our SAMRI [4] suite).
Session 2
posteroanterior superoinferior reference tissue
ID
bregma skull 0.00 -0.25 bregma skull
lambda skull -3.97 0.07 bregma skull skull
1 -0.50 0.18 lambda skull skull
2 -1.00 0.27 lambda skull skull
3 -1.50 0.35 lambda skull skull
4 -2.00 0.43 lambda skull skull
5 -2.50 0.56 lambda skull skull
6 -3.00 0.70 lambda skull skull
7 -3.50 0.98 lambda skull skull
8 -4.00 1.54 lambda skull skull
9 -4.48 2.12 lambda skull skull
VTA -3.50 4.25 bregma skull brain
DR -0.60 3.40 lambda skull brain
Table 1: Example skullsweep data obtained by lowering a glass pipette to the
skull repeatedly, along a generous area surrounding the implant target. Note
that each skull point is assigned a reference; these nested references are itera-
tively resolved by StereotaXYZ, until internally all distances are relative to the
canonical template origin (bregma).
−15 −10 −5 0 5
Posteroanterior(bregma) [mm]
−8
−6
−4
−2
0
2
Inferosuperior(bregma) [mm]
45°Insertion — Leftright(bregma) = 0.00mm
Skull
Target
Incision [PA/IS=-7.12/-0.67mm]
Insertion [3.61mm]
Insertion Axis
Figure 4: The simplest plotting functionality of StereotaXYZ represents all features of interest on a clear Cartesian grid.
This plotting method requires no MRI data availability, and is generally only recommended if the automatically generated
targeting plots are to be augmented with manual drawing or calculation.
Insertion [3.61mm]
Skull
Target
Incision [LR/PA/IS=0.00/-7.12/-0.67mm]
YZ/XY=45/0°Insertion
Figure 5: Features of interest co-registered to a generic MRI template. This plotting form does not require any preliminary MRI session, and could thus be integrated into any operative workflow. Another
advantage of template usage is high resolution, which may help better identify structures visually. Of course, such a plot cannot inform targeting decisions regarding idiosyncrasies in individual animals’ neuroanatomy.
Insertion [3.61mm]
Skull
Target
Incision [LR/PA/IS=0.00/-7.12/-0.67mm]
YZ/XY=45/0°Insertion
Figure 6: Analogously to figure 5, the features of interest can be co-plotted on a previously obtained animal-specific MR image. This is a trade-off between resolution and apparent image quality, and
nauroanatomical information specifically representative of the animal at hand. If the only neuroanatomical variation is expected to be due to broadly generalizable factors such as age, it is possible to construct a
more specific template for an animal group.
Session 3
Intended DR Target
Postoperative Verification
Figure 7: A post-operation MR scan can serve to inspect the success of the implantation procedure in vivo, ahead of any experimental group assignment (or full exclusion) of the animal. In the case at hand,
the implant is accurately positioned along the left-right axis, but only modesty well aligned with the intended target along the rostrocaudal and inferosuperior axes. This is most likely due to downward skipping of
the insertion point on the sloped cranium (loading on both aforementioned axes on which there is imprecision). Such issues are best addressed via improvements in the stereotactic drill holder stability, and illustrate
that in no case should the drill procedure for such implantations be performed free-handed.
References
[1] Siddharth R. Vora, Esra D. Camci, and Timothy C. Cox.
Postnatal ontogeny of the cranial base and craniofacial skeleton in male
c57bl/6j mice: A reference standard for quantitative analysis.
Frontiers in Physiology, 6:417, 2016.
[2] Brian B. Avants, Nicholas J. Tustison, Michael Stauffer, Gang Song, Baohua
Wu, and James C. Gee.
The insight toolkit image registration framework.
Frontiers in Neuroinformatics, 8:44, 2014.
[3] Krzysztof Gorgolewski, Christopher D Burns, Cindee Madison, Dav Clark,
Yaroslav O Halchenko, Michael L Waskom, and Satrajit S Ghosh.
Nipype: a flexible, lightweight and extensible neuroimaging data processing
framework in python.
Front Neuroinform, 5, 08 2011.
[4] Horea-Ioan Ioanas, Markus Marks, Dominik Schmidt, Florian Aymanns, and
Markus Rudin.
Samri - small animal magnetic resonance imaging, November 2017.
www.aic-fmi.ethz.ch Swiss Society for Neuroscience Annual Meeting — February 9, 2018 ioanas@biomed.ee.ethz.ch