PDF (526 KB)

J Neurosurg 99:89–99, 2003

Localization of stimulating electrodes in patients with Parkinson disease by using a three-dimensional atlas–magnetic resonance imaging coregistration method JÉRÔME YELNIK, M.D., PHILIPPE DAMIER, M.D., PH.D., SOPHIE DEMERET, M.D., DAVID GERVAIS, M.D., ERIC BARDINET, PH.D., BOULOS-PAUL BEJJANI, M.D., CHANTAL FRANÇOIS, PH.D., JEAN-LUC HOUETO, M.D., ISABELLE ARNULF, M.D., DIDIER DORMONT, M.D., DAMIEN GALANAUD, M.D., BERNARD PIDOUX, M.D., PH.D., PHILIPPE CORNU, M.D., AND YVES AGID, M.D., PH.D. Institut National de la Santé et de la Recherche Médicale U289, Centre d’Investigation Clinique, Fédération de Neurologie, Services de Neuroradiologie, Neurochirurgie, et Neurophysiologie, Centre National de la Recherche Scientifique UPR640, Hôpital de la Salpêtrière, Paris; and Institut National de la Recherche en Informatique et en Automatique, Epidaure Project, Sophia Antipolis, France Object. The aim of this study was to correlate the clinical improvement in patients with Parkinson disease (PD) treated using deep brain stimulation (DBS) of the subthalamic nucleus (STN) with the precise anatomical localization of stimulating electrodes. Methods. Localization was determined by superimposing figures from an anatomical atlas with postoperative magnetic resonance (MR) images obtained in each patient. This approach was validated by an analysis of experimental and clinical MR images of the electrode, and the development of a three-dimensional (3D) atlas–MR imaging coregistration method. The PD motor score was assessed through two contacts for each of two electrodes implanted in 10 patients: the “therapeutic contact” and the “distant contact” (that is, the next but one to the therapeutic contact). Seventeen therapeutic contacts were located within or on the border of the STN, most of which were associated with significant improvement of the four PD symptoms tested. Therapeutic contacts located in other structures (zona incerta, lenticular fasciculus, or midbrain reticular formation) were also linked to a significant positive effect. Stimulation applied through distant contacts located in the STN improved symptoms of PD, whereas that delivered through distant contacts in the remaining structures had variable effects ranging from worsening of symptoms to their improvement. Conclusions. The authors have demonstrated that 3D atlas–MR imaging coregistration is a reliable method for the precise localization of DBS electrodes on postoperative MR images. In addition, they have confirmed that although the STN is the main target during DBS treatment for PD, stimulation of surrounding regions, particularly the zona incerta or the lenticular fasciculus, can also improve symptoms of PD.

KEY WORDS • deep brain stimulation • magnetic resonance imaging • basal ganglion

bilateral DBS of the STN has proved to be effective in the treatment of severe forms of PD,25 with cardinal motor symptoms being improved by 70 to 80% and the need for an antiparkinsonian drug regimen being decreased by 60 to 70%.8,20,24,28,40 The efficacy of STN stimulation is dependent on two conditions: strict selection of patients and accurate placement of electrodes. Patient selection, which is becoming progressively more standardized, is mainly based on a good response to levodopa treatment, whether accompanied by disabling dyskinesias or not, and on the absence of cognitive deficiencies and active psychiatric disorders.25,44 The accuracy of electrode placement depends primarily

C

ONTINUOUS

Abbreviations used in this paper: AC = anterior commissure; DBS = deep brain stimulation; LF = lenticular fasciculus; MR = magnetic resonance; PC = posterior commissure; PD = Parkinson disease; RN = red nucleus; STN = subthalamic nucleus; UPDRS = Unified PD Rating Scale; ZI = zona incerta; 3D = three-dimensional.

J. Neurosurg. / Volume 99 / July, 2003

on the reliability of the surgical procedure and on the criteria with which the target is selected. In this study, the STN was chosen as the optimal target on the basis of experimental data2,5,6 as well as experience derived from electrode localization in treated patients; in other words, the targeted region of the STN in a given patient was that which gave the best results in those who had already undergone the operation. Therefore, localization involves two intricate issues: preoperative localization of the target and postoperative localization of the electrodes. Different strategies based on ventriculography, MR imaging, or a combination of the two, together with intraoperative electrophysiological identification of the STN and clinical evaluation have been developed for preoperative target localization.3,8,20,23–25,28,40 In this study, the target was localized directly on MR images and confirmed intraoperatively by electrophysiological recording results.3 Postoperative electrode localization was performed by coregistration of figures from an anatomical atlas with MR images obtained in the patient, a method al89

J. Yelnik, et al. TABLE 1 Therapeutic effects of STN stimulation in 10 patients* Postoperative Condition

Preoperative Condition Case No. 0 1 2 3 4 5 6 7 8 9 mean SD

Age (yrs), Sex 65, M 50, M 64, F 61, F 40, M 46, F 56, M 45, M 60, M 54, M 54 3

Hoehn & Yahr Stage (off/on) 5/2.5 3/1.5 4/3 4/2 5/4 5/3 4/2.5 5/4 4/3 3/2 4.2 0.2/ 2.8 0.3

Motor Treatment Duration Hoehn & Yahr Motor Treatment Fluctuations Dyskinesia (mg/day) (mos)† Stage (off/on) Fluctuations Dyskinesia (mg/day) 4 3 6 4 5 5 3 5 5 5 4.5 0.3

7 1450 10 6 1550 12 10 1150 13 8 1750 4 4 1250 27 9 1500 6 6 1900 5 9 1550 17 9 2440 2 9 850 4 7.7 0.6 1540 140 10 2

2.5/1 1/0 1/0 3/1 3/1 1.5/1 1.5/0 1/0 1/0 0/0 1.6 0.3‡/ 0.4 0.2‡

0 0 0 0 0 0 0 0 2 0 0.2 0.2‡

2 650 1 400 0 0 4 500 2 1050 3 700 0 400 0 425 2 200 0 75 1.4 0.5‡ 440 100‡

* SD = standard deviation. † Number of months between surgery and postoperative clinical evaluation. ‡ The probability value is less than 0.05 when comparing postoperative (time of the study) with preoperative Hoehn and Yahr score, motor fluctuations (UPDRS, Part IV-A), dyskinesia (UPDRS, Part IV-B), and the dosage of antiparkinsonian drugs administered (expressed as an equivalent dose of levodopa).

ready applied during pallidal stimulation45 and now validated by an analysis of the reliability of MR signals. This approach was applied at our center in a series of 10 patients with PD who were successfully treated using STN DBS. Two contacts per hemisphere were tested in each patient, the “therapeutic contact” and a “distant contact,” the latter defined as the next but one to the therapeutic contact. Forty sites were thus tested, the results from which we evaluated the therapeutic efficacy of stimulation of the STN and surrounding structures. Clinical Material and Methods Patient Population Ten patients successfully treated at our center by using DBS of the STN were included in this study. The surgical procedure for the implantation of electrodes was based on preoperative stereotactic MR imaging, intraoperative microrecordings, and clinical assessment of the effects of high-frequency stimulation.20 On the day after electrode implantation, MR images were acquired to ensure the correct positioning of the electrodes and the absence of intracerebral bleeding. A few days later, programmable stimulators were implanted in the subclavicular area and connected to the electrodes by a tunneled subcutaneous extension cable. The electrodes used (model 3389; Medtronic, Inc., Minneapolis, MN) were quadripolar, thus permitting stimulation to be applied at four individual contacts numbered from 0 to 3, 3 being the uppermost and 0 the lowest. Postoperatively, the effects of stimulation were assessed through each of the four contacts of the right and left electrodes to determine the contacts through which STN stimulation best alleviated symptoms of PD without inducing permanent side effects. The stimulator was then fine-tuned to choose the optimal contact (or sometimes the two optimal contacts) and the electrical parameters (frequency, pulse width, and voltage) to be used continuously to treat the patient.3,20 At the time of the study, all patients showed significant improvement with an 80% reduction in motor fluctuations 90

(assessed using the UPDRS, Part IV-A), an 82% reduction in dyskinesias (UPDRS, Part IV-B),12 and a 74% reduction in antiparkinsonian drug treatment administered (expressed as an equivalent dose of levodopa) in comparison with the preoperative evaluation (Table 1). The study protocol was approved by the ethics committee of our hospital, and all patients gave their informed written consent. Analysis of the Geometrical Features of the Electrode and its Four Contacts on MR Images

In the present study and a previous one, postoperative localization of electrodes was based on MR imaging data.3,45 To confirm the reliability of this approach, we performed a detailed analysis of the characteristics of an MR image of the electrode. The proximal portion of the electrode (model 3389; Medtronic, Inc.) consists of four nickel conductor wires insulated within a polytetrafluoroethylene jacket tubing. Its terminal portion is composed of four metallic (platinium/iridium) noninsulated contacts at 0.5-mm intervals, which generate an artifact on the MR image. To determine the precise characteristics of this artifact, we performed an experiment to ascertain the relationship between the artifact and the electrode itself. A glass tube, the internal diameter of which exactly fit the external diameter of the electrode (1.27 mm), was positioned in a bowl filled with agarose frost. The bowl was fixed in the antenna of the 1.5-tesla MR imaging unit (General Electric, Milwaukee, WI) used for patients with PD and submitted to a T1-weighted sequence acquisition (Fig. 1). The electrode was then inserted into the glass tube in such a way that the inferior part of the electrode reached the inferior part of the glass tube, and a second T1-weighted MR imaging sequence was obtained. The bowl and the glass tube were kept in exactly the same position for the two acquisitions. The two MR images were analyzed using rigid transformation,32 reformatting, synchronization viewers, and extraction algorithms.27 The MR imaging aspect of the electrode was also examined in our series of patients. For each of them, a plane conJ. Neurosurg. / Volume 99 / July, 2003

Localization of stimulating electrodes: 3D atlas–MR imaging coregistration

FIG. 1. Analysis of the MR image of the electrode. A: Magnetic resonance images demonstrating a quadripolar electrode placed within a glass tube that is maintained in a bowl of agarose frost. Three orthogonal sections of the image are shown. B: Two orthogonal sections of the MR image of the glass tube obtained after reformatting. The glass tube is vertically oriented. Note the indentations present along the tube. C: Two orthogonal sections of the MR image demonstrating the electrode placed in the glass tube. Note that the proximal portion (P) is cylindrical, whereas the terminal portion (T) is deformed. The blue and red lines in B and C intersect at the same voxel in the two synchronized images. D: An MR image demonstrating the extracted contours of the glass tube, which are traced in yellow. E: An MR image of the tube contours is superimposed onto an MR image of the electrode. Note that the proximal portion of the electrode fits exactly the image of the glass tube, whereas its terminal portion is larger. F: A postoperative MR image obtained after bilateral implantation of quadripolar electrodes into the STN. The proximal (P) and terminal (T) portions of the electrode are indicated (blue arrows). The MR image was obtained in an obliquely oriented slice containing the entire trajectory of the left electrode. The cylindrical hypointense signal at the tip of the electrode corresponds to the artifact induced by the four metallic contacts. G: An MR image of the right electrode in the same patient. Four indentations corresponding to the four contacts are indicated (blue arrows). H: An MR image displaying the left electrode in another patient. A 6-mm-long trace is superimposed on the terminal portion of the electrode. I: Schematic drawing of comparative dimensions (in millimeters) of the electrode and its artifact.


91

J. Yelnik, et al. of the four contacts, which was overlaid on this MR image (Figs. 1 and 2A). Anatomical Localization of Contacts on Postoperative MR Image Acquisition

FIG. 2. Contact identification and reformatting of the postoperative MR acquisition. A: An oblique MR image superimposed with a scaled template of the electrode in the prolongation of the axis of the proximal portion. The four consecutive contacts (0, 1, 2, and 3) are identified. B: A sagittal MR image revealing a section of the AC and PC points from which the horizontal plane is defined. The AC–PC line made a 38˚ angle with the axial acquisition plane. Four horizontal slices (1, 11, 21, and 31) separated by 1 cm are indicated. C: Axial MR image demonstrating Slice 1, which is located 20 mm above the AC–PC plane. The arrow indicates the proximal portion of the electrode. D: Axial MR image revealing Slice 11 located 10 mm above the AC–PC plane, with an arrow indicating the proximal portion of the electrode. E: Axial MR image exhibiting Slice 21 located at the AC–PC level, with an arrow indicating the upper part of the terminal portion of the contacts. F: Axial MR image demonstrating Slice 31 located 10 mm below the AC–PC level, with an arrow indicating the lowest contact of the electrode. Note that the proximal portion of the electrode is thinner than the terminal contacts.

taining the entire trajectory of the electrode, obliquely oriented with reference to both the frontal and sagittal planes, was extracted from the MR image acquisition (Fig. 1) and the main characteristics of the electrode were analyzed. Individual contacts were identified using a scaled template 92

In this study, we used the Voxtool software (General Electric Europe, Buc, France) run on an Advantage Windows Workstation,10 which provides a cursor that can be visualized simultaneously in three orthogonal planes, with its 3D coordinates being displayed. This system was used to determine the coordinates of each contact as well as any relevant point on the MR image. Anatomical localization of each contact was then performed using two different methods: coordinate measurement and 3D atlas–MR imaging registration. The coordinate measurement method was based on determining the coordinates of the stereotactic landmarks, namely, the AC, the PC, and a point within the interhemispheric cerebral fissure. The three coordinates of each identified contact were also measured in the same coordinate system and transformed into coordinates based on the AC– PC line by applying a previously published transformation matrix.10 These three coordinates were then transferred onto figures from an atlas based on the AC–PC line,33 and the localization of the contact, expressed in terms of its position within or outside the STN, was thus determined. The 3D atlas–MR imaging coregistration procedure was devised to create a scaled fusion of the contours of the atlas figures33 and the MR image obtained in each patient. The first step consisted of reformatting the postoperative MR image acquisition to obtain slices parallel to the ventricular AC–PC system (Fig. 2). Three stereotactic landmarks (namely, the AC, the PC, and a point on the interhemispheric cerebral fissure) were selected to define the midsagittal plane and its two orthogonal planes (that is, the axial or horizontal plane and the coronal or frontal plane). Slices (1mm thick) were then obtained in the horizontal plane (Fig. 2). The second step involved registration of atlas sections onto MR imaging slices. The correspondence between the series of MR imaging slices and the atlas sections was determined using landmarks that were clearly identifiable and had precise ventral and dorsal limits. The most stable landmarks were found to be the dorsal borders of the putamen and of the optic tract. The sections in the atlas,33 previously digitized using locally developed software, were then deformed by applying independent linear scaling factors along the anteroposterior and mediolateral axes (see Appendix; Fig. 3). The best fit between the MR imaging slice and an atlas section was based on the superimposition of specific landmarks: the lateral wall of the third ventricle, the anterior columns of the fornix, the mamillothalamic tract, the medial and lateral borders of the putamen, the external contour of the mesencephalon, and the AC and PC commissures. Each contact, identified as described earlier, could then be localized within a precise anatomical structure. Assessment of the Clinical Effects of STN Stimulation on Parkinsonian Symptoms

In each patient the effects of stimulation were assessed through two contacts of each electrode implanted in the left and right hemispheres. One of the contacts, referred to as the therapeutic contact, was the one through which STN J. Neurosurg. / Volume 99 / July, 2003


FIG. 3. A: The atlas section PC 3.5 superimposed onto a reformatted MR image of the corresponding section in a patient. Both images are at the same magnification, but the atlas contours do not fit the MR image. B: The figure from the atlas has been enlarged along the mediolateral and anteroposterior axes so that the landmarks coincide: AC, medial border of the putamen (Pu), anterior column of the fornix (Fx), and mamillothalamic tract (MTT). Note that the artifact of the contact now appears within the STN.

stimulation was continuously applied to treat the patient (the contact known from the immediate postoperative assessment to be the most effective in alleviating symptoms of PD was chosen in the event of two or more contacts being used continuously). The other contact, referred to as the distant contact, was separated from the therapeutic one by one contact. Clinical assessment was performed in the morning, the patient having been deprived of antiparkinsonian drugs for longer than 12 hours and bilateral stimulation for 30 minutes. A 2.5-V stimulation (frequency 130 Hz, pulse width 60 sec) was applied for 15 minutes through each of the four analyzed contacts. These four periods of stimulation plus a 15-minute no-stimulation condition were applied in a counterbalanced order by a study nurse, with the investigator (S.D.) and the patient being blinded to the condition. A 5-minute rest period was observed between each of the testing conditions. Patient assessment was based on the observation of any significant clinical changes and a motor examination (UPDRS, Part III) performed in the last 5 minutes of the application of each of the conditions. The UPDRS was used to define subscores for left- and right-sided akinesias (Items 23–26), rigidity (Item 22), and tremor (Items 20–21) as well as an axial subscore (Items 18, 19, and 22 for neck rigidity and Items 27–30). Results Analysis of the Geometrical Features of the Electrode and its Four Contacts on MR Images

The relationship between the MR imaging artifact generated by the electrode and its actual geometrical features was analyzed by comparing an MR image of the electrode within a glass tube and an MR image of the same glass tube alone. As the tube, which was maintained in a bowl filled with agarose frost (Fig. 1A), was obliquely oriented with respect to the acquisition planes, the two images were reformatted to allow visualization of the entire tube (Fig. 1B and C). First, the two images were aligned with each other by applying a rigid transformation. The resulting transformaJ. Neurosurg. / Volume 99 / July, 2003

tion was found to be identical, thus confirming that the glass tube was maintained in exactly the same position for the two acquisitions. The 3D images of the bowl in the two acquisitions thus being exactly superimposed, we were able to compare the position of the actual electrode (given by the tube position in the first image) with its MR image (second image). The two reformatted images were examined using synchronized viewers in which a 3D cursor can pinpoint simultaneously the same voxel in the two registered volumes. Visual inspection showed that the center of the glass tube was localized at the center of the artifact (red and blue lines in Fig. 1B and C). Two different portions could be identified: a proximal portion consisting of a cylindrical hyposignal and a terminal portion consisting of a larger and more irregular hyposignal (Fig. 1C). The glass tube as well as the proximal portion of the electrode had small (1.5 to 1.7 mm), regularly spaced indentations along its border and a hypersignal on one side. The width of the hyposignal was the same for the electrode and the glass tube (3.7 mm), but the hypersignal was more intense for the electrode. The voxels corresponding to the glass tube were then extracted from the first image (Fig. 1D) and superimposed on the image of the electrode (Fig. 1E). As demonstrated in these two figures, the image of the tube fits exactly the hyposignal of the proximal portion of the electrode (Fig. 1E). Conversely, the terminal portion of the electrode was larger than the diameter of the tube. The in vivo aspect of the electrode, examined on postoperative MR images, was very similar in the 10 patients with PD. The electrode consisted of two distinct portions when viewed in a plane containing its entire trajectory. The proximal portion (Fig. 1F) appeared as a cylindrical hyposignal with small (1.5–1.7 mm), regularly spaced indentations along its entire trajectory. The width of this hyposignal varied from 1.7 to 1.9 mm, that is, slightly more than the thickness of the electrode (1.27 mm). A hypersignal was observed at the border of the hyposignal (Fig. 1G and H), but because the hyposignal exactly fit the actual electrode (Fig. 1E), the electrode was identified solely on the basis of the hyposignal. The terminal portion of the electrode (Fig. 1F) 93

J. Yelnik, et al.

FIG. 5. Images demonstrating localization of 20 distant contacts determined using 3D atlas–MR imaging registration in each of 10 patients. Same presentation as in Fig. 4.

FIG. 4. Images demonstrating the localization of 20 therapeutic contacts determined using 3D atlas–MR imaging coregistration in each of 10 patients. Each contact is identified by the patient number (from 0 to 9). The mediolateral and rostrocaudal dimensions are indicated by the two centimetric orthogonal axes, the origin of which corresponds to the PC point. The dorsoventral dimension is given for each level by its position with reference to the PC plane.

was larger than its proximal portion in all patients: the width was 3.5 0.7 mm and the length was 10.7 1.1 mm (Fig. 1I). In some cases four indentations could be clearly identified (Fig. 1G), with a 6-mm distance between the uppermost and lowest indentations, corresponding to the actual distance between the center of the dorsal and ventral contacts (Fig. 1I). In other cases the terminal portion was cylindrical (Fig. 1F) or more irregular (Fig. 1H). In all cases, individual contacts were identified using a scaled template of the four contacts which was overlaid on this MR image (Figs. 1H and I and 2A). The template was superimposed on the electrode so that it was parallel to the long axis of its proximal portion and centered in its terminal portion. Anatomical Localization of Contacts on Postoperative MR Image Acquisition

Contact localization was first determined by measuring coordinates based on the AC–PC line and transferring them to figures from an atlas.33 Using this method, the 20 therapeutic contacts appeared to be localized between the dorsoventral levels AC–PC 1 and AC–PC 6, with 20% (four 94

of 20) being localized within the STN, 30% (six of 20) on the border of the STN, and 50% (10 of 20) outside the STN. Using the 3D atlas–MR imaging registration method, in which atlas contours are fitted to the particular dimensions of the brain in each patient, the localization of the therapeutic contacts was restricted to dorsoventral levels 1.5 to 4.5 of the STN (Fig. 4), with 75% (15 of 20) within the STN, 10% (two of 20) on the border of the STN, and 15% (three of 20) outside the STN. The three therapeutic contacts located outside the STN were in the ZI, the LF, and the midbrain reticular formation (that is, the region dorsal and lateral to the RN).29 The distant contacts were localized between dorsoventral levels AC–PC 1 and AC–PC 9.5 in different structures: six in the STN, one in the ZI, one in the LF, one in the RN, four in the SN, and seven in the midbrain reticular formation (Fig. 5). Of the 40 contacts tested (20 therapeutic and 20 distant contacts), 21 were localized in the STN, two at the border of the STN, eight in the midbrain reticular formation, four in the SN, two in the ZI, two in the LF, and one in the RN. Assessment of the Clinical Effects of STN Stimulation on Parkinsonian Symptoms

The effects of stimulation through therapeutic contacts on symptoms of PD are presented in Figs. 6 and 7. All 40 contacts tested are represented as squares that are approximately positioned in the figure according to their anatomical location. Squares (contacts) located in a given structure are grouped together. The four contacts located at the border of the STN are included in the STN group. The 20 therapeutic contacts are indicated by a bold underlining of characters within the squares. Seventeen therapeutic contacts (85%) were located within or on the border of the STN. Stimulation applied through most of these contacts markedly improved ( 25%) the four analyzed symptoms of PD: 94% were associated with improved akinesia, 100% with improved rigidity, 87% with improved tremor, and J. Neurosurg. / Volume 99 / July, 2003


FIG. 6. Diagram showing the localization of the 40 contacts tested (either therapeutic or distant) and the clinical effect of their stimulation on akinesia. Each square corresponds to a contact identified in the patient (Cases 0–9), the right (R) or left (L) side, and the number of the contact (0–3). Therapeutic contacts are indicated by bold underlining. Squares are placed according to their actual dorsoventral position, from dorsal level PC 1 to ventral level PC 9.5. They are arranged approximately from left to right according to their medial or lateral position. The contacts localized in a given region are grouped together. The clinical effect on akinesia, ranging from 50% improvement (that is, worsening) to 100% improvement, is indicated by a colored scale. MRF = midbrain reticular formation.

88% with improved axial signs. Stimulation delivered through 10 contacts improved the four analyzed symptoms, through five contacts improved three symptoms, through one contact improved two symptoms (akinesia and axial signs), and through one contact improved one symptom (akinesia). One contact was associated with worsened akinesia and two contacts with worsened axial signs. Three therapeutic contacts were located in other structures. A contact located in the ZI (patient in Case 0, right contact 3) was linked to significant improvement ( 50%) of the four studied symptoms. One contact located in the LF (patient in Case 9, left contact 3) was associated with a 25% or more improvement in the four symptoms. Stimulation applied through one contact located in the midbrain reticular formation (patient in Case 8, right contact 2) improved the four analyzed symptoms by 25% or more. In summary, stimulation delivered through most therapeutic contacts located in the STN significantly improved the four analyzed symptoms of PD. The effect of stimulation applied to three therapeutic contacts located in other structures, namely, the ZI, LF, and midbrain reticular formation, was also significantly positive. The effects of stimulation through the distant contacts were more variable. Six distant contacts were located in the STN. Of these six, five were implanted in patients whose J. Neurosurg. / Volume 99 / July, 2003

FIG. 7. Diagram showing the localization of the 40 contacts tested (either therapeutic or distant) and the clinical effect of their stimulation on rigidity. Same presentation as in Fig. 6.

therapeutic and distant contacts were both located in the STN. Of these five, one was associated with four improved symptoms, two with three improved symptoms, one with two improved symptoms, and one with one improved symptom. The sixth distant contact, located in the STN (patient in Case 9, left contact 1), was associated with an improvement ( 25%) in the four symptoms. In this same patient, the therapeutic contact was located in the LF (left contact 3) and was associated with an effect similar to that of the distant contact. The 14 remaining distant contacts were located in different regions. One distant contact was located in the ZI, and stimulation applied through this contact worsened axial signs but improved akinesia and tremor (patient in Case 5, left contact 3). Another distant contact was located in the LF (patient in Case 7, right contact 3); stimulation delivered through this contact worsened akinesia but improved other symptoms. Seven contacts were located in the midbrain reticular formation. These contacts were associated with variable effects ranging from a 50% worsening to a 75% improvement. Two of them (patient in Case 9, right contact 0; patient in Case 6, left contact 1) were related to a 25% or more improvement in the four symptoms. One contact located in the RN was linked to very weak effects (0–25%) on the four symptoms. Four contacts were located in the SN and were associated with variable effects ranging from a 50% worsening to a 75% improvement of symptoms. In summary, stimulation applied through distant contacts improved parkinsonian symptoms when the contacts were located in the STN. Contacts located in other structures were associated with variable effects ranging from worsening to improvement of symptoms, but the use of some had a positive effect on all four symptoms. 95

J. Yelnik, et al. Discussion The efficacy of STN stimulation is crucially dependent on the accurate placement of the therapeutic contact, which in turn depends on two additional steps: preoperative localization of the target and postoperative localization of the electrode. Different approaches have been used to localize the target in a given patient. Currently, ventriculography is often considered to be the gold standard for target determination,4,26,34 even though data from comparative studies have shown that computerized tomography scanning17,39 and MR imaging are at least as reliable as ventriculography.7,34,36 Magnetic resonance imaging is used without ventriculography at our center,10,20 but we, like many other groups, also use intraoperative electrophysiological recording and clinical testing to confirm target localization.16,20,21, 25,35 The postoperative localization of the electrode was the main subject of the present study. Geometrical Features of the Electrode and its Four Contacts on MR Images

In our study, contact identification and localization were both based on MR imaging studies. This presupposes that MR imaging provides a reliable representation of brain anatomy, a hypothesis that we have already verified. By using a 1.5-tesla unit for MR acquisition and the Voxtool software (General Electric Europe), run on an Advantage Windows workstation, a precision of 1 mm or less is obtained,10,11 which is sufficiently accurate to localize anatomical landmarks that were used to superimpose sections from an atlas with MR images. Conversely, local distortions on MR images are generated by the physical constitution of the electrode. The proximal portion of the electrode, a homogeneous structure with double-insulated metallic components, appeared on MR images as a cylindrical hyposignal with a diameter (1.7–1.9 mm) only slightly larger than the actual width of the electrode (1.27 mm). The indentations that were visible as well as the hypersignal that appeared along the hyposignal were due to a complex MR interaction between the different components of the electrode, a subject that is beyond the scope of this report. Note, however, that the glass tube used in the experiment induced similar artifacts (Fig. 1B), indicating that such artifacts are caused mainly by sudden changes of medium (for example, the electrode compared with nervous tissue in patients or glass compared with agarose frost in the experiment). The terminal portion of the electrode is conversely a heterogeneous structure in which the metallic components are not insulated. Thus, an MR image of its terminal portion was much more distorted than that of its proximal portion (Fig. 1). Nevertheless, the longitudinal axis of the proximal portion provides a reliable indicator of its actual axis and can therefore be used to trace the axis of the terminal portion. The localization of the contacts can then be determined by placing a 6-mm template of the contacts (Fig. 1I) at the longitudinal center of the MR image (Fig. 1H). Finally, we believe that an MR image of the electrode and its four contacts provides a reliable image of the actual electrode and contacts as long as certain precautions are taken, including extraction of the oblique axis of the proximal portion of the electrode and the use of a scaled template of the contacts. 96

Anatomical Localization of the Electrodes on Postoperative MR Image Acquisition

Given that the STN and adjacent structures are not discernible on T1-weighted MR images, a spin-echo T2weighted sequence could have been used, as performed at the preoperative step.3 This sequence is optimized for highest sensitivity to magnetic susceptibility, however, which increases the size of the electrode artifact and, in our experience, prevents the precise delineation of the STN. Anatomical localization thus requires the use of anatomical atlases. The simplest method of relating patient data to atlas contours is to measure the coordinates (based on the AC– PC system) of the patient’s electrodes and to transfer them onto the sections from the atlas. This method does not include any normalization procedure to correct for atlas–patient brain-size variations, however. In our series, such variations led to only 20% of contacts being located within the STN, even though the intended preoperative target was, in fact, the STN and its localization was confirmed intraoperatively by performing electrode recording. A second method involves correcting size variations by using the AC–PC length. The validity of this landmark, which can be identified on both ventriculography and MR imaging, has long been recognized.37 With this method, however, one presupposes that size variations are equivalent along the anteroposterior, mediolateral, and dorsoventral axes, a supposition that has been disproved.30 A better method is to correct size variations using different normalization factors along the three axes, but this is not possible with ventriculography alone because no reliable ventricular landmark is available along the mediolateral axis. The main characteristic of the 3D atlas–MR imaging registration method is that figures in the atlas are registered using specific scaling along each axis. A first scaling is determined along the dorsoventral axis by the correspondence established between atlas figure sections and MR image sections. Scaling along the other two axes (the mediolateral and anteroposterior axes in our study) is obtained by independent linear deformation of atlas figure sections along each axis. Another characteristic of the 3D atlas–MR imaging registration method is that registration is performed by adjusting all the contours of cerebral regions rather than a limited number of discrete landmarks. The accuracy of registration was also improved by selecting local landmarks13 instead of distant cortical landmarks, such as those proposed in the proportional grid system of Talairach and Tournoux.38 Postoperative localization of the electrodes has rarely been specifically investigated based on the literature. In the reported studies MR imaging is often performed a few days after electrode implantation to check their location, but the precise technique of localization is not described.22,24,25 In fact, MR imaging is commonly reported to induce spatial distortions, thus requiring intraoperative electrophysiological recording and clinical testing instead of anatomical targeting.35 In a study of patients with PD who had been treated with pallidotomy,15 lesion location was determined postoperatively on volumetric reformatted MR imaging, as was used in the present study. A comparison with conventional (midcommissural point and intercommissural line) landmarks demonstrated that anatomical landmarks identifiable on J. Neurosurg. / Volume 99 / July, 2003

Localization of stimulating electrodes: 3D atlas–MR imaging coregistration MR images were more relevant than the traditional ventricular landmarks. In another study,46 researchers used three anatomical targeting methods that each revealed significant differences in the localization of the STN compared with its electrophysiologically based localization. These authors concluded that anatomical targeting was not an accurate enough procedure. The atlases used in the anatomical targeting were scaled to the mean AC–PC distance in the 15 patients studied, which does not account for either interindividual variations or different scaling along the three anatomical axes. In a recent study,43 a combination of several techniques was used to localize the STN target, including computerized tomography scanning, MR imaging (T1- and T2weighted), ventriculography, and teleradiography. The final position of the electrode was documented in the operating room by using teleradiography, these data being transferred onto preoperative planning MR images and intraoperative ventriculography images. The position of the electrode was also expressed in relation to the AC–PC line and transferred onto figures from an atlas.33 It is extremely difficult to make a precise comparison between the procedures used in different institutions because the accuracy of the procedures is highly dependent on the quality of the imaging devices used and on the experience of the team. The combination of methods used by Voges, et al.,43 produced good clinical results but could be difficult to implement at other institutions lacking their experienced personnel, thus increasing the risk of error to the detriment of the reliability of the methods. Finally, we believe that MR imaging coupled with atlas coregistration is a reliable technique for visualizing intracerebral structures as long as certain methodological precautions are respected, namely MR reformatting, oblique identification of electrodes, and 3D multiscaled registration. The future development of the 3D atlas–MR imaging coregistration method includes its application for the preoperative localization of the target, the development of a new atlas specifically devoted to 3D registration (the atlas of Schaltenbrand and Wharen33 used in this and many other studies contains too few, irregularly spaced sections), and the verification of this method by comparison with microelectrode recordings and other targeting methods. It is hoped that this method will become sufficiently standardized to be useful at many different centers. Assessment of the Clinical Effects of STN Stimulation on Parkinsonian Symptoms

Most of the contacts analyzed in this study were located within the STN. All patients analyzed had at least one contact located within each STN, except for two (Cases 0 and 8) in whom the right electrode was slightly medial to the STN. Stimulation applied through most of the contacts located within the STN, whether therapeutic or distant, had a positive effect (25–100% improvement) on akinesia (91%; Fig. 6), rigidity (82%; Fig. 7), tremor (84%), and axial signs (83%). Stimulation delivered through the remaining contacts had negative effects on parkinsonian symptoms despite contact localization within the STN. This could be due to different factors. Although an error in the determination of the localization of the contacts seems unlikely; the atlas used33 was not devised for 3D registration and could therefore be a source of some imprecision. Another reason could J. Neurosurg. / Volume 99 / July, 2003

be that STN stimulation was either inactive or less active in some patients. It is important to note that stimulation through a given contact may act on akinesia but not rigidity (Case 4, left contact 3), or akinesia but not on tremor (Case 7, right contact 1). This dissociation suggests that DBS of the STN does not simply produce a global inactivation of the STN and the output nuclei of the basal ganglia, but has a more complex effect on circuitry, which in turn could be affected differently in different patients. Some of the contacts located outside the STN were found to be associated with improved signs of PD (for example, Case 8, right contact 2; Case 6, left contact 1 in the midbrain reticular formation; Case 0, right contact 3 in the ZI; and Case 9, left contact 3 in the LF). This suggests that these regions could be involved in the pathophysiology of parkinsonian symptoms. The reticular area between the RN and the SN is a complex region containing disseminated dopaminergic cell bodies of Area A8 as well as different axonal bundles (brachium conjunctivum, nigral efferent axons, serotoninergic ascending axons from the raphe nuclei, and cholinergic axons from the pedunculopontine nucleus). Although their involvement in parkinsonian symptoms is unknown, a clinical benefit could result from the activation or inactivation of these structures. A beneficial effect of DBS on the ZI is more plausible because activation of this structure has been demonstrated both in patients with PD41 and in a rat model of the disease.31,42 A beneficial effect of DBS on the LF could be related to inactivation of pallidothalamic fibers that run in this bundle, a hypothesis that has received recent support.43 Parkinson Disease: Where is the Functional Target of High-Frequency Stimulation?

The subthalamic area is a complex region consisting of the STN, situated at the geometrical center of this region, and different gray nuclei surrounded by several fiber bundles. Because of the geometrical features of the electrode used and the shape and dimensions of the STN, we were able to test therapeutic and distant contacts located in both the STN and surrounding structures. The STN appeared to be the optimal target given that 85% of the therapeutic contacts were located within this structure. This result confirms that the STN is by far the most suitable target for neurosurgical treatment of PD, a finding consistent with the current models of basal ganglia circuitry,1,9 experimental studies in rats and monkeys,6,14,18 and clinical results reported by different groups.3,4,8,20,24,25,28,40 Stimulation of the ZI was found to significantly improve parkinsonian symptoms in one of the patients analyzed (Case 0). This case was unique, however, and we cannot propose this structure as a reliable target. Nevertheless, it is worth noting that the ZI is hyperactive in PD,31 indicating that it could be involved in the pathophysiology of parkinsonism. Two other therapeutic contacts were located in surrounding regions, namely in the LF and the midbrain reticular region. This might be explained by the passage of pallidothalamic axons that run in the LF, as recently proposed,43 or by fibers from the pedunculopontine nucleus that ascend in the midbrain reticular region. Data from this study confirms that the theoretical target is the STN, but also indicates that adjacent structures such as the ZI or the LF could be alternative targets. 97

J. Yelnik, et al. Conclusions Magnetic resonance imaging is a powerful tool that allows the 3D structure of the brain in living patients to be examined in detail. Our first concern was to verify the reliability of the MR signal generated by DBS electrodes in patients with PD. Thus, we devised an experiment in which the physical position of an electrode could be precisely related to its MR image. Results demonstrated that the electrode on the MR image, although it appeared larger than the size of the actual electrode, was stable from one patient to another and that the localization of the electrode could be precisely deduced from this image. Our second concern was to determine with great accuracy the anatomical localization of the electrodes in a given patient. To this end, we developed a method by which figures from an anatomical atlas can be coregistered with the MR image obtained in each patient by using independent scaling factors along the three axes. This method, referred to as “3D atlas–MR imaging coregistration,” was used to localize 40 stimulating contacts in 10 patients with PD. This approach confirmed the STN as the main target for DBS in the treatment of PD, but showed that stimulation of surrounding regions, particularly the ZI or the LF, could also improve symptoms of PD. Further studies are needed, however, to improve the 3D atlas–MR imaging coregistration method, including the development of an atlas better suited to 3D registration and its application in preoperative targeting. Appendix The 3D atlas–MR imaging coregistration procedure was performed as follows. Digitized sections of the atlas33 were imported as separate images and rendered transparent in a Microsoft PowerPoint software file. Slice MR images in each patient were imported into another PowerPoint file. Atlas sections were transferred from the atlas file to the patient file by using the copy and paste functions and were deformed by using the format/image/size function of the software with the proportional option deactivated. The same height and width scale percentages were applied to all atlas section/MR slice pairs for a given patient. A copy of the PowerPoint atlas file used in our study can be obtained on request from Dr. Yelnick. Readers are invited to contact him for advice on implementing the procedure at their own center. Acknowledgments The nurses of the Centre d’Investigation Clinique who cared for the patients are gratefully acknowledged. References 1. Alexander GE, Crutcher MD: Functional architecture of basal ganglia circuits: neural substrates of parallel processing. Trends Neurosci 13:266–271, 1990 2. Aziz TZ, Peggs D, Sambrook MA, et al: Lesion of the subthalamic nucleus for the alleviation of 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP)-induced parkinsonism in the primate. Mov Disord 6:288–292, 1991 3. Bejjani BP, Dormont D, Pidoux B, et al: Bilateral subthalamic stimulation for Parkinson’s disease by using three-dimensional stereotactic magnetic resonance imaging and electrophysiological guidance. J Neurosurg 92:615–625, 2000 4. 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 55 (Suppl 6): S40–S44, 2000

98

5. Benazzouz A, Gross C, Féger J, et al: Reversal of rigidity and improvement of motor performance by subthalamic high-frequency stimulation in MPTP-treated monkeys. Eur J Neurosci 5: 382–389, 1993 6. Bergman H, Wichmann T, DeLong MR: Reversal of experimental parkinsonism by lesions of the subthalamic nucleus. Science 249:1436–1438, 1990 7. Bourgeois G, Magnin M, Morel A, et al: Accuracy of MRI-guided stereotactic thalamic functional neurosurgery. Neuroradiology 41:636–645, 1999 8. The Deep-Brain Stimulation for Parkinson’s Disease Study Group: Deep-brain stimulation of the subthalamic nucleus or the pars interna of the globus pallidus in Parkinson’s disease. N Engl J Med 345:956–963, 2001 9. DeLong MR: Primate models of movement disorders of basal ganglia origin. Trends Neurosci 13:281–285, 1990 10. Dormont D, Cornu P, Pidoux B, et al: Chronic thalamic stimulation with three-dimensional MR stereotactic guidance. AJNR 18: 1093–1107, 1997 11. Dormont D, Zerah M, Cornu P, et al: A technique of measuring the precision of an MR-guided stereotaxic installation using anatomic specimens. AJNR 15:365–371, 1994 12. Fahn S, Elton RL, Members of the UPDRS Development Committee: Unified Parkinson’s Disease Rating Scale, in Fahn S, Marsden CD, Goldstein M, et al (eds): Recent Developments in Parkinson’s Disease. Florham Park, NJ: Macmillan Healthcare Information, 1987, Vol 2, pp 153–163 13. François C, Yelnik J, Percheron G: A stereotaxic atlas of the basal ganglia in macaques. Brain Res Bull 41:151–158, 1996 14. Gao DM, Benazzouz A, Piallat B, et al: High-frequency stimulation of the subthalamic nucleus suppresses experimental resting tremor in the monkey. Neuroscience 88:201–212, 1999 15. Gross RE, Lombardi WJ, Hutchison WD, et al: Variability in lesion location after microelectrode-guided pallidotomy for Parkinson’s disease: anatomical, physiological, and technical factors that determine lesion distribution. J Neurosurg 90:468–477, 1999 16. Guridi J, Rodriguez-Oroz MC, Lozano AM, et al: Targeting the basal ganglia for deep brain stimulation in Parkinson’s disease. Neurology 55 (Suppl 6):S21–S28, 2000 17. Hariz MI, Bergenheim AT: A comparative study on ventriculographic and computerized tomography-guided determinations of brain targets in functional stereotaxis. J Neurosurg 73:565–571, 1990 18. Hirsch EC, Perier C, Orieux G, et al: Metabolic effects of nigrostriatal denervation in basal ganglia. Trends Neurosci 23:S78–S85, 2000 19. Hoehn MM, Yahr MD: Parkinsonism: onset, progression and mortality. Neurology 17:427–442, 1967 20. Houeto JL, Damier P, Bejjani PB, et al: Subthalamic stimulation in Parkinson disease: a multidisciplinary approach. Arch Neurol 57:461–465, 2000 21. Hutchison WD, Allan RJ, Opitz H, et al: Neurophysiological identification of the subthalamic nucleus in surgery for Parkinson’s disease. Ann Neurol 44:622–628, 1998 22. Krack P, Pollak P, Limousin P, et al: Subthalamic nucleus or internal pallidal stimulation in young onset Parkinson’s disease. Brain 121:451–457, 1998 23. Kumar R, Lang AE, Rodriguez-Oroz MC, et al: Deep brain stimulation of the globus pallidus pars interna in advanced Parkinson’s disease. Neurology 55 (Suppl 6):S34–S39, 2000 24. Kumar R, Lozano AM, Sime E, et al: Comparative effects of unilateral and bilateral subthalamic nucleus deep brain stimulation. Neurology 53:561–566, 1999 25. Limousin P, Krack P, Pollak P, et al: Electrical stimulation of the subthalamic nucleus in advanced Parkinson’s disease. N Engl J Med 339:1105–1111, 1998 26. Limousin P, Pollak P, Benazzouz A, et al: Bilateral subthalamic nucleus stimulation for severe Parkinson’s disease. Mov Disord 10:672–674, 1995


Localization of stimulating electrodes: 3D atlas–MR imaging coregistration 27. Lorenson WE, Cline HE: Marching cubes: a high resolution 3D surface construction algorithm. Comput Graph 21:163–169, 1987 28. Olanow CW, Brin MF, Obeso JA: The role of deep brain stimulation as a surgical treatment for Parkinson’s disease. Neurology 55 (Suppl 6):S60–S66, 2000 29. Parent A: Carpenter’s Human Neuroanatomy, ed 9. Baltimore: Williams & Wilkins, 1996 30. Percheron G: Ventricular landmarks for thalamic stereotaxy in Macaca. J Med Primatol 4:217–244, 1975 31. Perier C, Vila M, Féger J, et al: Functional activity of zona incerta neurons is altered after nigrostriatal denervation in hemiparkinsonian rats. Exp Neurol 162:215–224, 2000 32. Roche A, Malandain G, Pennec X, et al: The correlation ratio as a new similarity measure for multimodal image registration. Lect Notes Comput Sci 1496:1115–1124, 1998 33. Schaltenbrand G, Wahren W: Atlas for Stereotaxy of the Human Brain, ed 2. Stuttgart: Thieme, 1977 34. Schuurman PR, de Bie RM, Majoie CB, et al: A prospective comparison between three-dimensional magnetic resonance imaging and ventriculography for target-coordinate determination in frame-based functional stereotactic neurosurgery. J Neurosurg 91:911–914, 1999 35. Starr PA, Vitek JL, Bakay RAE: Ablative surgery and deep brain stimulation for Parkinson’s disease. Neurosurgery 43:989–1015, 1998 36. Starr PA, Vitek JL, DeLong M, et al: Magnetic resonance imaging-based stereotactic localization of the globus pallidus and subthalamic nucleus. Neurosurgery 44:303–314, 1999 37. Talairach J, David M, Tournoux P, et al: Atlas d’Anatomie Stéréotaxique: Repérage Radiologique Indirect des Noyaux Gris Centraux des Régions Mésencéphalo-Sous-Optique et Hypothalamique de L’homme. Paris: Masson, 1957 38. Talairach J, Tournoux P: Co-Planar Stereotaxic Atlas of the Human Brain: 3-Dimensional Proportional System: An Approach to Cerebral Imaging. New York: Thieme Medical Publishers, 1988 39. Tasker RR, Dostrovsky JO, Dolan EJ: Computerized tomography


40. 41.

42.

43.

44. 45.

46.

(CT) is just as accurate as ventriculography for functional stereotactic thalamotomy. Stereotact Funct Neurosurg 57:157–166, 1991 Valldeoriola F, Pilleri M, Tolosa E, et al: Bilateral subthalamic stimulation monotherapy in advanced Parkinson’s disease: longterm follow-up of patients. Mov Disord 17:125–132, 2002 Vila M, Levy R, Herrero MT, et al: Consequences of nigrostriatal denervation on the functioning of the basal ganglia in human and nonhuman primates: an in situ hybridization study of cytochrome oxidase subunit I mRNA. J Neurosci 17:765–773, 1997 Vila M, Perier C, Féger J, et al: Evolution of changes in neuronal activity in the subthalamic nucleus of rats with unilateral lesion of the substantia nigra assessed by metabolic and electrophysiological measurements. Eur J Neurosci 12:337–344, 2000 Voges J, Volkmann J, Allert N, et al: Bilateral high-frequency stimulation in the subthalamic nucleus for the treatment of Parkinson disease: correlation of therapeutic effect with anatomical electrode position. J Neurosurg 96:269–279, 2002 Welter ML, Houeto JL, Tezenas du Montcel S, et al: Clinical predictive factors of subthalamic stimulation in Parkinson’s disease. Brain 125:575–583, 2002 Yelnik J, Damier P, Bejjani BP, et al: Functional mapping of the human globus pallidus: contrasting effect of stimulation in the internal and external pallidum in Parkinson’s disease. Neuroscience 101:77–87, 2000 Zonenshayn M, Rezai AR, Mogilner AY, et al: Comparison of anatomic and neurophysiological methods for subthalamic nucleus targeting. Neurosurgery 47:282–294, 2000

Manuscript received November 11, 2002. Accepted in final form March 28, 2003. The study was supported by INSERM and the National Parkinson Foundation-Miami. Address reprint requests to: Jerome Yelnik, M.D., INSERM U289, Hôpital de la Salpêtrière, 47, boulevard de l’Hôpital, 75013 Paris, France. email: [email protected].

99