Surgical Neurology 63 (2005) 357 – 363

www.surgicalneurology-online.com

Comparison of 2-dimensional magnetic resonance imaging and

3-planar reconstruction methods for targeting the
subthalamic nucleus in Parkinson disease
Yuri M. Andrade-Souza, MD, Jason M. Schwalb, MD, Clement Hamani, MD, PhD,
Tasnuva Hoque, BSc, Jean Saint-Cyr, PhD, Andres M. Lozano, MD, PhD*
Division of Neurosurgery, Department of Surgery, University of Toronto and University Health Network, Toronto, Ontario, Canada M5T 2S8
Received 19 January 2004; accepted 10 May 2004

Abstract Objective: The study aims to compare 2-dimensional (2D) and 3-planar (3P) reconstruction
magnetic resonance imaging (MRI) methods of targeting the optimal region of the subthalamic
nucleus (STN) for chronic stimulation in patients with Parkinson disease.
Methods: We studied 14 patients with Parkinson disease treated with bilateral STN deep brain
stimulation (DBS) (28 STN targets). Electrode implantation was based on direct and indirect
targeting based upon the position of the anterior and posterior commissures using 2D MRI, with
selection of the final target based on microelectrode recording. Optimal settings, including the
contacts used, were determined during the clinical follow-up. The position of the best contact was
defined with postoperative MRI. Optimal contact position was compared to targets calculated by the
direct method from the preoperative 2D MRI and 3P reconstruction. Optimal contact position was
also compared to the indirect targets calculated from the preoperative 2D MRI and 3P reconstruction.
The distance between the targets and the position of the best contact were calculated.
Results: The mean improvement in OFF-period Unified Parkinson Disease Rating Scale III
subscores with STN DBS was 52%. The mean distance between the optimal contact position and the
direct target was 4.66 mm (SD = 1.33) using the 2D MRI and 3.49 mm (SD = 1.29) using the 3P
reconstruction (t test, P b .001). The mean distance between the optimal contact and the indirect
target was 3.42 mm (SD = 1.34) using the 2D MRI and 2.61 mm (SD = 0.97; t test, P = .001) using
the 3P reconstruction. The variance of the direct target was less using the 3P reconstruction than
using the 2D MRI (F test, P = .002), indicating greater precision. Similarly, the variance of the
indirect target using the 3P reconstruction was less than using the 2D MRI (F test, P = .012).
Conclusion: Indirect and direct targets chosen using 3P reconstruction more closely approximate the
position of the clinically optimal contact than targets chosen using 2D MRI.
D 2005 Elsevier Inc. All rights reserved.
Keywords: Deep brain stimulation; Subthalamic nucleus; Parkinson disease; Neurosurgery; Magnetic resonance imaging;
Stereotactic surgery

* Corresponding author. Division of Neurosurgery, Toronto Western
Hospital, Toronto, Ontario, Canada M5T 2S8. Tel.: +1 416 603 6200; fax:
+1 416 603 5298.
E-mail addresses: yurisouzamd@hotmail.com (Y.M. Andrade-Souza)8
lozano@uhnres.utoronto.ca (A.M. Lozano).
0090-3019/$ – see front matter D 2005 Elsevier Inc. All rights reserved.
doi:10.1016/j.surneu.2004.05.033
1. Introduction
Despite the relatively small size, ovoid shape, and
oblique disposition of the subthalamic nucleus (STN)
[17,21], it can be targeted for the treatment of movement
disorders with excellent results [9,11]. Ventriculography [2],
computerized tomography [1,7,22], and magnetic resonance
imaging (MRI) [16,19] have all been used for operative
358 Y.M. Andrade-Souza et al. / Surgical Neurology 63 (2005) 357 – 363
targeting. Several studies have demonstrated that MRI is as
accurate as the other methods [1,3,10]. With the progress of
the computer-assisted surgery, 3-planar (3P) reconstruction
of the stereotactic images has been proposed as a way to
increase the reliability and efficacy of this procedure
[3,4,18] by correcting for frame placement and minimizing
MRI-related distortion.
In this study, we compared conventional 2-dimensional
(2D) MRI with 3P reconstruction methods of targeting the
STN in patients with Parkinson disease. We compared
modified direct targets (3 mm above the center of the
nucleus), derived from either the 2D MRI or the 3P
reconstruction, with the coordinates of the contact giving
the optimal clinical result postoperatively. Indirect targets
based on the coordinates of the anterior commissure (AC)
and posterior commissure (PC), using either 2D MRI or 3P
reconstruction, were also compared to the optimal contact
position.
2. Patients and methods
We reviewed 14 patients with Parkinson disease treated
with bilateral subthalamic deep brain stimulation (28
subthalamic targets). For inclusion in this study, a preoper-
ative stereotactic MRI and a postoperative MRI (1-5 days
postoperative) were required. The parameters used for both
pre- and postoperative MRIs (1.5-T unit, Signa Model;
General Electric Medical System, Milwaukee, WI) are
shown in Table 1. The techniques used for deep brain
stimulation electrode insertion are described elsewhere [8].
The x-coordinate was defined as the lateral-medial distance,
the y-coordinate as the anterior-posterior distance, and the z-
coordinate as the superior-inferior distance. The target of the
first microelectrode track was based upon the 2D MRI–
based indirect target for x- and z-coordinates and the mean
of the y-coordinates from the 2D MRI–based direct and
indirect targets. No correction was made for uneven frame
placement before choosing the target for the first pass of the
microelectrode.
The postoperative MRI was used to target the final
position of the electrode tip (Model 3387 quadripolar lead)
as well as the medial-lateral and anterior-posterior trajectory
Table 1
MRI specifications
Parameter T2-weighted sequence
Image mode 2D
TR (ms) 4000.0
TE (ms) 90.0
Echo train 8 (Leksell series G; Elekta Instruments, Atlanta, GA) were
No. of slices 20
Matrix size 256  256
Slice thickness (mm) 2
Slice gap (mm) 0 mean rod marking error was calculated by the software and
Signal means 3 registered. Coronal and sagittal images were reconstructed
Bandwidth (kHz) 3.29
Field of view (mm) 270
angles. The coordinates of the optimal contact were
calculated as previously described, using the contact that
gave the optimal clinical results at follow-up, the position of
the tip, and the trajectory angles [16]. These coordinates
were derived relative to the AC and PC in Talairach space. If
the patient had better results using double monopolar
stimulation, the midpoint between the 2 negative contacts
was used as the final target. If bipolar stimulation was
optimal, the negative contact was defined as the final target
because this is the area of highest stimulation (charge is
concentrated around the cathode).
2.1. Indirect and direct 2D targets
Both direct and indirect targeting from 2D MRI were
used for surgical planning. A standard T2-weighted axial 2D
MRI was used to target the posterior border of the AC and
the anterior border of the PC. The indirect target was
defined as a point 12 mm lateral, 3 mm posterior, and 3 mm
inferior to the AC-PC midpoint (MCP). The 2D direct target
was calculated using coronal sections from the 2D T2-
weighted MRI. The STN was identified as a hypointense,
ovoid structure, lateral to the most anterior part of the red
nucleus and superior to the substantia nigra. The center of
the STN was targeted and its coordinates calculated.
Preliminary analysis (data not shown) revealed that this
target was far from the position of the optimal contact. This
was not unexpected because the optimal target for chronic
stimulation within the STN appears to lie near the superior
portion of the nucleus [6,13,16,23]. Therefore, the z-
coordinate (superior-inferior) of the direct target was
modified superiorly by 3 mm to obtain closer coordinates
to the optimal region for stimulation. This modified direct
target was closer to the optimal contact position than the
direct target (data not shown).
For our post hoc analysis, frame-based coordinates were
transformed into AC-PC coordinates as a way to correct for
frame placement and to compare different patients. Uneven
frame placement was measured in terms of the roll
(displacement in the coronal plane), yaw (displacement in
the axial plane), and pitch (displacement in the sagittal plane
relative to the AC-PC line) to analyze its contribution to
errors in targeting.
2.2. Indirect and direct 3P target
Axial T2-weighted fast-spin echo sequence images were
transferred to the StealthStation. Using the FrameLink
software (FrameLink 4.1, StealthStation Mach 4.1; Med-
tronic SNT, Minneapolis, MN), the fiducials of the frame
recognized, transforming the entire image volume into
stereotactic coordinate space. During this procedure, the
based on the information from the axial images. The AC and
PC were targeted using the 3 planes. Two points in the
 Y.M. Andrade-Souza et al. / Surgical Neurology 63 (2005) 357 – 363 359

Table 2 indirect target and the 3P reconstruction–based indirect

Frame displacement (roll, yaw, and pitch, in degrees) in the 2D images and target [2]. Because distances are continuous variables and
mean rod marking error (mm) registered by the software in the 3P
reconstruction (n = 14)
we were performing comparisons by pairs, Student t test
was used to determine significance. Variance was calculated
Mean Minimum Maximum
as the square of the SD and compared using the F test. A P
Roll 2.07 0 5.00
value of less than .05 was selected as statistically significant.
Yaw 2.75 0 6.34
Pitch 7.90 0 19.80 We also tried to correlate the mean rod marking error with
Rod marking error 1.17 0.32 1.89 the target error in the 3P reconstruction.

midline (sylvian aqueduct and falx cerebri) were identified 3. Results

to define the AC-PC plane. This plane contains the AC-PC The mean improvement of the patients on Unified
line and is perpendicular to a plane containing the sylvian Parkinson Disease Rating Scale III was 52% (SD = 11.93,
aqueduct, falx cerebri, AC, and PC. Defining the AC-PC minimum = 40, and maximum = 71), comparing the
plane corrects for frame placement but keeps the image in preoperative bOFFQ medication and the postoperative bONQ
stereotactic coordinate space, allowing direct identification stimulation bOFFQ medication, as previously reported [9].
of any point in the image volume. The mean, minimum, and maximum frame placement errors,
After 3P reconstruction, the indirect target was derived in addition to the rod marking error, are shown in Table 2.
using the same constant distances (12 mm lateral, 3 mm We were unable to find any correlation between the rod
posterior, and 3 mm inferior) from the MCP, but relative to marking error and the distance of the 3P direct and indirect
the AC-PC plane. The direct target was defined in a similar targets from the optimal contact position (R = 0.09, P = .64;
fashion as from the 2D MRI, but using the 3P images R = 0.2, P = .21). The mean distance between the MCP
simultaneously. After the superior region of the STN was registered in the 2D MRI and in the 3P reconstruction was
targeted, the software displayed both the frame-based and 1.00 mm (SD = 0.62, minimum = 0 and maximum = 1.73).
the AC-PC–based coordinates. However, to compare the 3P There was a normal distribution of distances between the
reconstruction–based targets and the 2D MRI–based targets, optimal contact and the targets obtained with each method.
we transformed the frame-based coordinates displayed in The individual x-, y-, and z-coordinate differences between
the software to AC-PC–based coordinates based on the AC- the targets and the optimal contact position are shown in
PC coordinates targeted in the 2D MRI. This method was Table 3. Except for the x-distance between the direct target
chosen as a way to obtain consistent coordinates for the AC and the optimal contact, 3P-based targets were always more
and PC because anatomically based targets in 2D MRI and accurate than the 2D-based targets. However, the variance
3P-reconstructed MRI do not necessarily have the same of the x-distance for the 3P-based direct target was smaller
coordinates. The AC-PC distance was calculated in both the than that for the 2D-based direct target, suggesting greater
2D MRI and the 3P reconstruction. precision with the 3P-based method.
2.3. Statistical analysis As shown in Table 4, the mean Euclidean distances
between the targets and the optimal contact positions (mean
The stereotactic coordinates of the AC, PC, MCP, direct, error of the target) were statistically smaller for the 3P-based
and indirect targets were entered into a computerized targets than for the 2D-based targets, signifying greater
spreadsheet (SPSS, version 11.5) for statistical analysis. accuracy when using the 3P reconstruction. Variances of the
The roll, yaw, and pitch of the frame and the mean rod
marking error were also entered.
Differences between the coordinates obtained by our Table 3
Difference between the target coordinates and position of the optimal
various targeting methods and the final postoperative contact (error of the target) in millimeters along the individual axes using
position of the optimal contact were calculated as Euclidean 2D MRI and 3P reconstruction
distances in xyz space. The Euclidean distance is the Method Coordinate Mean Minimum Maximum SD Variance
straight-line distance between 2 points (p1 and p2). In a
Direct 2D x 1.57 0.00 6.50 1.37 1.90
xyz system (3-dimensional) with p1 at (x 1, y 1, z 1) and p2 at y 2.44 0.00 6.00 1.86 3.48
(x 2, y 2, z 2), it is M[(x 1 x 2)2 + ( y 1 y 2)2 + (z 1 z 2)2]. z 2.75 0.00 6.00 1.53 2.36
Differences along the individual x-, y-, and z-axes were also Direct 3P x 1.76 0.00 5.00 1.10 1.23
analyzed to determine if there was a greater error in one y 1.98 0.00 5.00 1.30 1.69
z 1.66 0.00 4.00 1.10 1.22
direction than the others.
Indirect x 1.44 0.00 3.50 0.95 0.91
The Kolmogorov-Smirnov test was used to determine if 2D y 2.10 0.00 5.00 1.51 2.28
the distances fit a normal distribution. Two null hypotheses z 1.51 0.00 4.50 1.28 1.63
tested were: there is no difference between the 2D MRI– Indirect x 1.23 0.00 3.50 0.83 0.69
based direct target and the 3P reconstruction–based direct 3P y 1.58 0.50 3.00 0.72 0.52
z 1.28 0.00 3.00 0.94 0.89
target [1]; there is no difference between the 2D MRI–based
360 Y.M. Andrade-Souza et al. / Surgical Neurology 63 (2005) 357 – 363

Table 4 between the targets and the optimal contacts, with 95%
Distance between the target coordinates and optimal contact position (error confidence intervals.
of the target) with statistical analysis
Method Mean Minimum Maximum SD Variance
Direct 2D 4.66 2.44 8.90 1.33 1.77 4. Discussion
Direct 3P 3.49 1.50 6.48 1.29 1.66
Our statistical analysis shows greater accuracy and
Indirect 2D 3.42 1.11 5.91 1.34 1.80
Indirect 3P 2.61 1.11 4.69 0.97 0.95 precision in targeting with 3P reconstruction than with 2D
MRI. This study agrees with previous studies that suggested
Method Method the value of 3P reconstruction–based targeting. Slavin and
Analysis of the means a colleagues [18] reported that they used fewer micro- or
P Direct 2D Direct 3P b.001b macroelectrode recording tracts after instituting the use of
Indirect 2D Indirect 3P .001b 3P-based targeting, although it is unclear if this was
statistically significant. Even so, it may not be valid to
Analysis of the variancesc
P Direct 2D Direct 3P .002b attribute their decreased number of cannula passes to the
Indirect 2D Indirect 3P .012b change in targeting method because they used historical
a
Using the paired Student t test. controls. It may have been an improvement in some other
b
Statistically significant. aspect of their technique associated with experience. Cuny
c
Using the F test. and colleagues [3] compared 2D MRI–based direct target-
ing, ventriculography-based indirect targeting, and 3P
3P-based targets were also smaller than the 2D targets, reconstruction–based indirect targeting with the optimal
confirming the higher precision of the former (precision a 1/
variance). Fig. 1 graphically represents the mean distances

Fig. 1. The mean distance between the targets and the optimal contact Fig. 2. The direct and indirect targets, as well as the positions of the optimal
position (error of the target) with 95% confidence intervals. A, Direct target. contacts, using the 2D MRI and 3P reconstruction (sagittal view),
B, Indirect target. superimposed upon the STN from the Schaltenbrand and Wharen [17] atlas.
 Y.M. Andrade-Souza et al. / Surgical Neurology 63 (2005) 357 – 363
position for postoperative stimulation. They found that 3P
reconstruction–based targeting was the most precise tech-
nique for STN targeting. However, they did not test direct
targeting using the 3P reconstruction or indirect targeting
using 2D MRI. Thus, the differences they detected may
have been due to differences between direct and indirect
techniques rather than differences between 2D and 3P
methods.
The mean distance between the optimal contact and the
direct target was 4.66 mm (SD = 1.33) using the 2D MRI
and 3.49 mm (SD = 1.29) using the 3P reconstruction ( P b
.001). One reasonable explanation for this difference is that
the 2D direct target was registered in the coronal T2-
weighted MRI, which has more distortion than axial
sequences [14,15]. The 3P reconstruction uses the axial
images to construct the coronal and sagittal images. The
reconstructed coronal images are not obtained from a true
coronal MRI sequence and are therefore not subject to the
same distortion as the T2 MRI coronal sequence. In
addition, by reformatting the images orthogonally to
midsagittal plane, a more symmetrical subthalamus is seen.
Visualization of the sagittal, coronal, and axial planes at the
same time may also facilitate registration of the direct
targets (Fig. 2).
The fact that the most accurate result was obtained with
the 3P-based indirect target may be due to the elimination of
problems related to frame placement with this method. The
use of ear bars can decrease the roll and yaw on the 2D MRI
images. However, their use is not routine and they do not
reduce pitch error. Displacement secondary to the frame was
not corrected when calculating the indirect target from the
2D MRI. When using the 3P reconstruction, the software
automatically reformats the data parallel to AC-PC plane
and perpendicular to the true midsagittal plane. The
correction for the displacement in the 3 planes (roll, yaw,
and pitch) can be done using trigonometric calculations, as
described by other authors [5,12], but is very time
consuming.
We cannot conclude that the indirect target is better than
the direct target based on this study. The x- and z-
coordinates of the first microelectrode track were based
upon the 2D MRI–based indirect target. The y-coordinate of
the first track was the mean of the 2D MRI–based direct and
indirect targets. Thus, the indirect target could bias the final
position of the electrode. However, because our choice of
initial targets was based on the 2D MRI direct and indirect
targets and not the 3P reconstruction, our retrospective
finding of greater accuracy using the 3P method is not
subject to this bias.
However, all of these methods have some degree of
inaccuracy. Indirect 3P-based targeting, which produced the
coordinates closest to the optimal contact position, still had
a mean error of 2.61 mm (1.11-4.69) when compared to the
position of the final optimal contact. Therefore, we
recommend the use of electrophysiology in determining
the final electrode position.
361
Finally, there was no correlation between the mean rod
marking error and target error, in agreement with the
findings of Steinmeier and colleagues [20] using a frameless
system.
5. Conclusion
The use of 2D MRI– and 3P reconstruction–based
targeting methods were compared in this study. We
compared both direct and indirect targets registered using
these 2 methods with the final position of the optimal
contact after clinical follow-up for 6 months. The use of the
3P-reconstructed images improved the accuracy and preci-
sion of both the modified direct and indirect STN targets in
patients with Parkinson disease.
References
[1] Alterman RL, Kall BA, Cohen H, et al. Stereotactic ventrolateral
thalamotomy: is ventriculography necessary? Neurosurgery
1995;37(4):717 - 21 [discussion 721-2].
[2] Benabid AL, Krack PP, Benazzouz A, et al. Deep brain stimulation of
the subthalamic nucleus for Parkinson’s disease: methodologic aspects
and clinical criteria. Neurology 2000;55(12 Suppl. 6):S40-4.
[3] Cuny E, Guehl D, Burbaud P, et al. Lack of agreement between direct
magnetic resonance imaging and statistical determination of a
subthalamic target: the role of electrophysiological guidance. J
Neurosurg 2002;97(3):591 - 7.
[4] diPierro CG, Francel PC, Jackson TR, et al. Optimizing accuracy in
magnetic resonance imaging-guided stereotaxis: a technique with
validation based on the anterior commissure-posterior commissure
line. J Neurosurg 1999;90(1):94 - 100.
[5] Dormont D, Cornu P, Pidoux B, et al. Chronic thalamic stimulation
with three-dimensional MR stereotactic guidance. AJNR Am J
Neuroradiol 1997;18(6):1093 - 107.
[6] Hamel W, Fietzek U, Morsnowski A, et al. Deep brain stimulation of the
subthalamic nucleus in Parkinson’s disease: evaluation of active
electrode contacts. J Neurol Neurosurg Psychiatry 2003;74(8):1036 - 46.
[7] Hariz MI, Bergenheim AT. A comparative study on ventriculographic
and computerized tomography-guided determinations of brain targets
in functional stereotaxis. J Neurosurg 1990;73(4):565 - 71.
[8] Hutchison WD, Allan RJ, Opitz H, et al. Neurophysiological
identification of the subthalamic nucleus in surgery for Parkinson’s
disease. Ann Neurol 1998;44(4):622 - 8.
[9] Kleiner-Fisman G, Fisman DN, Sime E, et al. Long-term follow up of
bilateral deep brain stimulation of the subthalamic nucleus in patients
with advanced Parkinson disease. J Neurosurg 2003;99(3):489 - 95.
[10] Kondziolka D, Dempsey PK, Lunsford LD, et al. A comparison
between magnetic resonance imaging and computed tomography for
stereotactic coordinate determination. Neurosurgery 1992;30(3):402 - 6
[discussion 406-7].
[11] Krack P, Batir A, Van Blercom N, et al. Five-year follow-up of
bilateral stimulation of the subthalamic nucleus in advanced Parkin-
son’s disease. N Engl J Med 2003;349(20):1925 - 34.
[12] Krauss JK, King DE, Grossman RG. Alignment correction algorithm
for transformation of stereotactic anterior commissure/posterior
commissure-based coordinates into frame coordinates for image-
guided functional neurosurgery. Neurosurgery 1998;42(4):806 - 11
[discussion 811-2].
[13] Lanotte MM, Rizzone M, Bergamasco B, et al. Deep brain stimulation
of the subthalamic nucleus: anatomical, neurophysiological, and
outcome correlations with the effects of stimulation. J Neurol
Neurosurg Psychiatry 2002;72(1):53 - 8.
362 Y.M. Andrade-Souza et al. / Surgical Neurology 63 (2005) 357 – 363
[14] Mack A, Czempiel H, Kreiner HJ, et al. Quality assurance in
stereotactic space. A system test for verifying the accuracy of aim in
radiosurgery. Med Phys 2002;29(4):561 - 8.
[15] Piovan E, Zampieri PG, Alessandrini F, et al. Quality assessment of
magnetic resonance stereotactic localization for Gamma Knife radio-
surgery. Stereotact Funct Neurosurg 1995;64(Suppl. 1):228 - 32.
[16] Saint-Cyr JA, Hoque T, Pereira LC, et al. Localization of clinically
effective stimulating electrodes in the human subthalamic nucleus on
magnetic resonance imaging. J Neurosurg 2002;97(5):1152 - 66.
[17] Schaltenbrand G, Wahren W. Atlas for stereotaxy of the human brain.
Stuttgart7 Thieme; 1977.
[18] Slavin KV, Anderson GJ, Burchiel KJ. Comparison of three
techniques for calculation of target coordinates in functional
stereotactic procedures. Stereotact Funct Neurosurg 1999;72:192 - 5.
[19] Starr PA, Christine CW, Theodosopoulos PV, et al. Implantation of
deep brain stimulators into the subthalamic nucleus: technical
approach and magnetic resonance imaging-verified lead locations. J
Neurosurg 2002;97(2):370 - 87.
[20] Steinmeier R, Rachinger J, Kaus M, et al. Factors influencing the
application accuracy of neuronavigation system. Stereotact Funct
Neurosurg 2000;75(4):188 - 202.
[21] Talairach J, David M, Tournoux P. Atlas d’Anatomie Stereotaxique.
Reperage Radiologique Indirect des Noyaux Gris Centraux des
Regions Mesencephalo-sous-optique et Hypothalmique de l’Homme.
Paris7 Masson; 1957.
[22] Vilela Filho O, da Silva DJ. Unilateral subthalamic nucleus lesioning:
a safe and effective treatment for Parkinson’s disease. Arq Neuro-
psiquiatr 2002;60(4):935 - 48.
[23] Yelnik J, Damier P, Demeret S, et al. Localization of stimulating
electrodes in patients with Parkinson disease by using a three-
dimensional atlas-magnetic resonance imaging coregistration method.
J Neurosurg 2003;99(1):89 - 99.
Commentary
In this paper, the authors compare the accuracy of 2D
MRI with 3P MRI for treatment planning during placement
of bilateral deep brain stimulator electrodes into the
subthalamic nucleus in patients with Parkinson disease.
Specifically, they identified a direct target (based upon
imaging of the STN itself) and an indirect target (based
upon stereotactic coordinates) using both 2D and 3P
imaging methods. The electrode pass(es) were based upon
the 2D direct and indirect targeting. Postoperative imaging
revealed the location of the most clinically effective
electrode. This location was then compared to the location
as predicted by 2D (direct and indirect) and 3P (direct and
indirect) imaging methods. The accuracy with which 2D
direct targeting matched up with the actual effective
electrode was compared with the accuracy of the 3P direct
targeting. A similar analysis was done to compare 2D and
3P indirect targeting. In this manner, the differences in
predictive value of 2D vs 3P could be illustrated in contrast
to previous work that mixed direct and indirect methods (ie,
an bapples to applesQ comparison). The data in this report
suggest that 3P preoperative planning is more accurate than
2D planning during both direct and indirect targeting
procedures.
This study provides yet another step in the technical
refinement of STN targeting, an obviously complex
endeavor. The next logical step is to complete a randomized
clinical trial to compare 2D with 3P targeting methods. It
may turn out that the increased precision (if real) obtained
with 3P targeting actually improves outcome from these
procedures. Of course, it is possible that this incrementally
increased accuracy does not affect outcome and is merely
splitting hairs at this point and not worth the time and effort.
Kim J. Burchiel, MD, FACS
Christopher J. Winfree, MD
Department of Neurosurgery
Oregon Health Sciences University
Portland, OR 97201, USA
Authors of this paper analyze their experience with MRI-
guided STN targeting and bring up several very interesting
points. First of all, they found that neither atlas-based
(indirect) nor visualization-based (direct) methods of target-
ing are universally effective in finding the optimal part of
STN. This, at least in part, may be because of the inadequate
accuracy of current anatomic visualization. In our practice,
we use (similar to the authors) MRI parameters in one
institution and higher-resolution imaging with 3-T scanner
in another. We found impressive improvement in visualiza-
tion of STN and overall surgical accuracy with smaller pixel
size [1] obtained with thinner cuts and higher image
resolution (512  512), although this could also be
explained by a better image contrast due to the higher
magnetic field.
Second, the paper once again shows the importance of
physiologic target confirmation. The error of up to 8.9 mm
may leave electrode far away from STN, and unless
physiologic testing (microrecording, micro- or macrostimu-
lation) is used, the results of such misplacement would
translate into lack of clinical improvement. Although the
paper does not provide a convincing argument to support
mandatory use of microelectrode recording, one may make
certain conclusions regarding precision of various physio-
logic modalities. In my opinion, accuracy provided by
microelectrode recording is much higher than that of
macrostimulation, and I feel more confident when I use it
even with better anatomic targeting techniques.
Also of importance, though not discussed by the authors
directly, is the point of old discussion of the site of best
stimulation location. Obviously, if stimulation of a certain
edge of STN or some place next to it provides better
symptomatic improvement, then we should adjust our
procedure by shifting the area we use for direct targeting
and changing those (atlas-based) coordinates that are used
for indirect targeting approach. After all (and most clinicians
agree), the atlases are based on a bnormalQ brain that may
indeed be different from the brains of patients with
Parkinson disease.
The only thing that was not brought up in this paper is that
both direct and indirect targeting still greatly depend on
manual delineation of STN and standard anatomic landmarks.