Presurgical mapping of language, motor and visual regions: initial experience and comparison to intra-operative stimulation.

Cynthia G. Wible, Daniel Kacher, Reisa Sperling, Juliana Pare-Blagoev, Arya Nabavi, Emmanouel Chatzidakis, Peter Black, William Wells, Ferenc Jolesz, Philip, E. Stieg, Ron Kikinis, Eben Alexander III, Robert W. McCarley

Harvard Medical School (Surgical Planning Laboratory) and Brockton VAMC, Boston MA.

I. Introduction

II. Methods

III. Conclusions

IV. References

V. Figures

I. INTRODUCTION

fMRI has been used at several cites for presurgical mapping of language and motor function and was found to have a high degree of sensitivity and specificity when compared to electrocortical stimulation (e.g. Fitzgerald et al., 1997).
Presurgical mapping of language, motor, and visual regions was done using several tasks at first in order to evaluate their usefulness.

Auditory and visual verb generation, the Boston Naming Test, category or letter word generation, word repetition, and listening to text being read. Control tasks used with some patients included listening to complex tone sequences, or to sequences of speech sounds (fragments of spoken phonemes that were put together to be the same length as words).

Hand Flexion.

Reversing checkerboard, a flashing light, coherent moving dots, and color stimuli.
fMRI results were visualized by the neurosurgeon during surgery along with 3D models of the patient’s tumor, brain, vessels, etc., and these were registered to the patient’s head throughout surgery. Neurosurgery was done either in the open MRI machine, the Magnetic Resonance Therapy site (MRT), or in a conventional operating room with a workstation and tracking equipment.
Diffusion Tensor Images were also taken for many patients; large white matter tracts can be seen on these images. Their usefulness for surgical planning is currently being evaluated.

 

II. METHODS

FUNCTIONAL MAGNETIC RESONANCE IMAGING (fMRI)

Overview. All images were acquired using a GE 1.5 Tesla Signa System with a HORIZON hardware/software package (GE Medical Systems, Milwaukee, WI).

Low-resolution anatomical images were taken using the same slice thickness and in the same location and plane as the functional images. The low-resolution 3D SPGR images were acquired and reformatted into 7mm contiguous coronal slices.

Functional images (EPIBOLD) were acquired in a continuous manner, with up to 102 whole brain acquisitions (the first 2 acquisitions were discarded); each set of 102 images was referred to as a functional experiment. At least 2 sets of acquisitions were taken per task. A gradient-echo echo-planar sequence was used to acquire 21 contiguous 7mm coronal slices of the whole brain with the following parameters: TE=40msec, TR=3 sec, FOV 24 cm, image resolution = 64 X 64, in-plane voxel edge = 3.75 mm.

Structural images were taken in a separate session and included a 124 slice SPGR and gadolinium enhanced angiography series.

Procedures for an fMRI Session:

General procedures. Subjects were placed in the magnet with the head held within an air-filled VAC-FIX (S&S X-Ray Products, Brooklyn, NY) cushion that inflated to conform to head shape to secure against movement. A personal computer (PC) equipped with a digital I/O board (National Instruments) was used for the behavioral task and to collect response information. A signal indicating the application of RF pulses by the functional pulse sequence was also fed into the computer and used to trigger the stimulus/response programs.

Auditory stimuli were stored as sound files on the computer and were presented presented with Labview programs using sound-insulated earphones (Silent Scan, Avotec, Jensen Beach, FL). Visual stimuli were computer-controlled and were back-projected onto a screen at the foot of the patient. Responses were collected using a hand held squeeze bulb that, when squeezed, sent a pulse of air through a tube to an air switch (PRES:AIR:TROL Corp, Mamaroneck, NY) that was connected to the computer.

fMRI data analyses. The mean of the pixel values of images in the 2 no-task conditions surrounding each task condition was computed to be compared with that task condition. T-tests were used to compare the average images for particular phases, pixel by pixel. T values that were above a threshold and that are surrounded by at least 3 other pixels with significant T values were considered active regions of interest. Cross-correlation tests were also used in the analyses, as well as those allowing for the visualization of the time-course responses of individual pixels. Results of all analyses were examined and final results of the cross-correlation analyses were usually used during surgery. Movement correction was done using the package AIR, however, there was very little movement in the data.

REGISTRATION OF FUNCTIONAL AND HIGH RESOLUTION ANATOMICAL DATA

The low-resolution MR anatomical scans acquired during the functional scanning session were automatically registered to the high resolution SPGR scans using the multi-step, iterative, MI algorithm (Wells et al., 1997; Wible et al., in preparation). The functional images were then transformed to the lattice of the high-resolution SPGR scans using the geometric transformation that was computed to move the low-resolution anatomical scans into the lattice of the high-resolution SPGR scans. The functional data were then reformatted into the lattice of the high-resolution scans and were written out as 1.5 mm slices. This step involved taking the pixels in the scans that were significantly activated and interpolating them using a KNN nearest neighbor algorithm.

This step allowed for the visualization during surgery of the fMRI data along with the 3D models of face/head, vessels, tumor, and brain.

CONVENTIONAL OPERATING ROOM PROCEDURES FOR THE REGISTRATION OF fMRI AND 3D ANATOMICAL MODELS TO THE PATIENT AND SURGICAL PROBE

 Skin-to-skin registration and LED tracking system.

Tracking light emitting diodes (LEDs) are first fixed to the patient's head (usually 5).

The patient's head contour was then registered to a 3D model of the skin derived from an MR scan. The brain and other tissues are also represented in this same model. During the skin-skin registration, a series of points from the patient's skin are recorded using an LED probe (Image Guided Technologies, Inc.). The information was tracked using an optical digitizer (Flashpoint 5000, Image Guided Technologies, Inc.).

The points recorded in real space are then matched to the 3D model using a two stage process. An initial rough alignment was obtained by recording the real-space location of three points, then manually matching those points on the MRI model. This initial registration was then refined by finding the optimal transformation that aligned all of the points onto the skin surface of the model.

During neurosurgery, the stimulating probe was tracked using LEDs.

Electrode grid locations were digitized by touching each one with a tracked probe and recording the location on the 3D models.

 

MAGNETIC RESONANCE THERAPY SITE (MRT) PROCEDURES FOR THE REGISTRATIONOF fMRI AND 3D ANATOMICAL MODELS TO THE PATIENT AND THE SURGICAL PROBE

fMRI and structural 3D models were registered to the patient’s head and visualized during surgery on an LCD monitor located near the magnet. This was done in much the same way as in the conventional OR, except that instead of skin contours, the brain of the patient was used for registration. First, a structural image of the patient was taken during just prior to surgery, this scan was then registered to the images from which the 3D models were computed using the MI algorithm. Probes (with attached LEDs) can be tracked using an optical tracking system (Flashpoint Model 5000, Image GuidedTechnologies, Boulder, CO) consisting of three infrared cameras mounted over the imaging volume. The probes and tracking system allow the operating physician to specify slice locations relative in position and orientation to the hand-held probe, and the same system can be used to mark points of stimulation during neurosurgery. At the time that these functional studies were done, the stimulating apparatus had not been acquired, but will be used in future procedures.

LANGUAGE MAPPING

Several language tests were used at first in order to determine their utility. Although temporal (Wernicke’s Area) and frontal (Broca’s Area and anterior prefrontal) activation was obtained in most subjects for most language tasks, individuals did vary in terms of which tests activated temporal, frontal, or both regions. Two tests, auditory and visual verb generation seemed to capture most of the language activity for the patients, a finding consistent with those of Fitzgerald et al.,1997. Other regions activated were visual regions (for visual language tasks), the cingulate, and the cerebellum. In over 11 patients tested on language tests, all but one showed at least some fMRI activity in temporal and frontal regions. One patient had extensive temporal/frontal damage with edema and only showed a few pixels of activation in the temporal lobe.

MOTOR MAPPING

The hand motor region was mapped in over ten patients. Four patients underwent intra-operative tests for sensory/motor function. Only one fMRI activation model did not show correspondence with intra-operative stimulation (shown below). In those patients not tested intra-operatively, the fMRI motor signal showed good localization with regard to conventional anatomic landmarks.

VISUAL MAPPING

Visual mapping was done in only one patient. All tests used in this patient were also tested in controls. The color fMRI test was the only one that did not show localized activation as expected (see McKeefry and Zeki, 1997; Sakai et al, 1995; Tootell and Taylor, 1995).

III. CONCLUSIONS

IV. REFERENCES

Binder JR, Frost JA, Hammeke TA, Cox RW, Rao SM, Prieto T. Human brain language areas identified by functional magnetic resonance imaging. Journal of Neuroscience 1997; 17 (1): 353-362.

Fitzgerald DB, Cosgrove GR, Ronner S, Jiang H, Buchbinder BR, Belliveau JW, Rosen BR, Benson RR. Location of language in the cortex: A comparison between functional MR imaging and electrocortical stimulation. AJNR 1997; 19: 1529-1539.

McKeefry DJ & Zeki S. The position and topography of the human colour centre as revealed by functional magnetic resonance imaging. Brain 1997; 120: 2229-2242.

Sakai K, Watanabe E, Onodera Y, Uchida I, Kato H, Yamamoto E, Koizumi H, Miyashita Y. Functional mapping of the human colour centre with echo-planar magnetic resonance imaging. Proc R Soc Lond 1995; 261: 89-98.

Tootell RBH & Taylor JB. Anatomical evidence for MT and additional cortical visual areas in humans. Cerebral Cortex 1995; 1: 39-55.

Wells WM, Viola P, Atsumi H, Nakajima S, Kikinis R. Multi-modal volume registration by maximization of mutual information. Medical Image Analysis 1996; 1 (1): 35-51.

Wible CG, Wells WM, Yoo SS, Kacher D, Kikinis R, Jolesz F, McCarley RW. The registration of fMRI and high resolution anatomical MRI scans using mutual information. In preparation.

V. FIGURES

Neurosurgeon: Dr. Black

N.H. is a 28 year old right handed male. He presented with a glioma with findings of astrocytic and oligodendrocytic origin located in the left parietal lobe.
N.H was tested using auditory and visual verb generation, word generation using categories, and word repetition, as well as hand motor movement during fMRI procedures. All language tests except word generation from categories produced activation in the left temporal regions, except that the activation from auditory verb generation was bilateral, with left activation being greater than right, and the activation from visual verb generation was in the right hemisphere. Auditory verb generation, visual verb generation, word generation from categories, and auditory word repetition all produced left frontal activation. Intraoperative Stimulation showed good co-localization with fMRI for motor function. Dr. Black conversed with the patient throughout surgery to ensure no speech difficulties. There were no postoperative deficits.
Neurosurgeon: Dr. Stieg

S.M. is a 29 year old right handed female who presented with a history of seizures and was found to have a large left temporal cavernous angioma.
S.M. was tested using auditory and visual verb generation, word generation using categories, and the Boston Naming Test. Activation for most tests was located in left temporal and frontal regions, except for the Boston Naming and visual verb generation, which produced right temporal and bilateral temporal activation, respectively.
Neurosurgeon: Dr. Stieg

J.T. is a young woman who presented with new onset seizures and a focal neurologic deficit on her left side and was admitted. She had facial nerve weakness as well as a mild left hemiparesis. An MRI scan was consistent with a low grade mass in the posterior frontal area.
J.T. performed a hand motor task during fMRI procedures. She was operated under general anesthesia. Electrophysiological phase reversal using subdural grid electrodes was used on the brain surface to determine the boundary between the motor and sensory cortices (center). Each point on the grid was probed and displayed on the 3D model. The results indicated a perfect correlation with the pre-operative 3D reconstruction and fMRI data.
Neurosurgeon: Dr. Stieg

D.T. is a 46 year old right-handed female who presented with new onset seizures and was found to have a large right frontoparietal arteriovenous malformation.
D.T. was tested with fMRI procedures during hand movement and the hand motor region was also localized using transcranial magnetic stimulation (TMS). Stimulation showed good correspondence with TMS and fMRI activation.
Neurosurgeon: Dr. Stieg

G.G. is a 46 year old right handed female who presented with probable seizures. MRI revealed a right occipital arteriovenous malformation.
G.G. was tested during fMRI using a reversing checkerboard, flashing black and white slides, coherent moving dots, and color Mondrians. The patient showed no visual deficits postoperatively.
Neurosurgeon: Dr. Alexander

J.J. is a 24 year old right handed female. She experienced episodes of numbness in her face, right hand, right torso and leg and also weakness in her right arm and leg. In the two months prior to diagnosis she experiences headaches, some severe. She was found to have a left frontal low grade glioma.
J.J. was tested using auditory and visual verb generation in addition to hand movement during fMRI procedures. She was the only patient tested thus far who showed bilateral activation in both frontal and temporal regions on both verb generation tests. The procedure was done in the MRT under general anesthesia. Dr. Alexander lifted the anesthesia several times throughout surgery to test her language abilities. She showed no language deficits postoperatively.