HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Schwartz, R. B. Articles by Jolesz, F. A. Search for Related Content PubMed PubMed Citation Articles by Schwartz, R. B. Articles by Jolesz, F. A.

Intraoperative MR Imaging Guidance for Intracranial Neurosurgery: Experience with the First 200 Cases¹

¹ From the Depts of Radiology (R.B.S., L.H., T.Z.W., D.F.K., A.A.Z., R.K., F.A.J.) and Surgery, Div of Neurosurgery (P.M.B., E.A., P.E.S., T.M.M., C.A.M.), Brigham and Women's Hospital, 75 Francis St, Boston, MA 02115. From the 1997 RSNA scientific assembly. Received Mar 18, 1998; revision requested May 14; revision received Aug 12; accepted Oct 26. Supported in part by NIH grants PO1 CA 67165, P41 RR13218, R01 RR11747, and R01 CA 45743 and by the National Science Foundation. Address reprint requests to R.B.S.

	Abstract

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 
PURPOSE: To review preliminary experience with an open-bore magnetic resonance (MR) imaging system for guidance in intracranial surgical procedures.

MATERIALS AND METHODS: A vertically oriented, open-configuration 0.5-T MR imager was housed in a sterile procedure room. Receive and transmit surface coils were wrapped around the patient's head, and images were displayed on monitors mounted in the gap of the magnet and visible to surgeons. During 2 years, 200 intracranial procedures were performed.

RESULTS: There were 111 craniotomies, 68 biopsies, 12 intracranial cyst evaluations, four subdural drainages, and five transsphenoidal pituitary resections performed with the intraoperative MR unit. In each case, the intraoperative MR system yielded satisfactory results by allowing the radiologist to guide surgeons toward lesions and to assist in treatment. In two patients, hyperacute hemorrhage was noted and removed. The duration of the procedure and the complication rate were similar to those of conventional surgery.

CONCLUSION: Intraoperative MR imaging was successfully implemented for a variety of intracranial procedures and provided continuous visual feedback, which can be helpful in all stages of neurosurgical intervention without affecting the duration of the procedure or the incidence of complications. This system has potential advantages over conventional frame-based and frameless stereotactic procedures with respect to the safety and effectiveness of neurosurgical interventions.

Index terms: Brain, MR, 10.121411, 10.121412, 10.12143 • Brain, surgery, 10.439 • Brain neoplasms, 10.146, 10.363, 10.37, 10.38 • Brain neoplasms, therapy, 10.1261, 10.1263, 10.1267 • Magnetic resonance (MR), guidance, 10.121411, 10.121412, 10. 12143 • Magnetic resonance, technology

	Introduction

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 
The goal of any intracranial neurosurgical intervention is the accurate localization, targeting, diagnosis, and treatment of intracranial abnormalities while inducing minimal damage to the intact brain. Frame-based and frameless stereotactic localization techniques, which make use of computed tomography (CT) or magnetic resonance (MR) imaging, are routinely used to guide surgeons to previously imaged and diagnosed brain lesions (1–6).

Although these techniques have enabled neurosurgeons to access, obtain biopsy specimens from, and, in selected cases, resect lesions with great efficiency, they have several constraints. First, the accuracy of conventional stereotactic surgery requires that preoperative landmarks conform to intraoperative anatomy, but the act of opening the skull and removing tissue may result in unpredictable distortions, shifts, and deformations of the exposed surgical site and surrounding structures. Second, with conventional preoperative imaging techniques, it is not possible to track the advancing biopsy needle (or other surgical instruments) as it passes through brain tissue, nor is it possible to correct the path once the trajectory is defined. Third, during open surgery, the boundaries of some lesions, in particular low-grade gliomas, may be difficult to determine by means of gross visual inspection or manipulation alone. Fourth, it usually is not possible intraoperatively to ascertain the potential communication between ventricles, cisterns, and other fluid-containing structures. Finally, after conventional surgery, surgical complications may not be evident for several hours, until the effects of anesthesia are reversed or until the late consequences of a developing edema or hemorrhage become clinically evident.

Owing to the need for an image-guidance method to address these issues, an intraoperative MR imaging system was devised (7). Through an industrial-academic collaboration, a vertically oriented, open-configuration, 0.5-T MR unit was developed to allow radiologists and surgeons to perform interventional or intraoperative procedures with direct MR imaging guidance. This system provides near real-time imaging, as well as navigational assistance, during surgical procedures performed in a sterile environment (8,9).

The feasibility of this method has been established at our institution for use in a variety of procedures (9), including neurosurgical applications (10,11). The continuous visual feedback provided by the system allows radiologists to monitor the accuracy of lesion targeting, determine the extent of resection and changes in anatomic relationships during the procedure, and identify any immediate intraoperative complications. In addition, the capabilities of the system have motivated radiologists to develop and use newer imaging techniques (12,13).

Several other groups have reported preliminary data on the use of interventional MR devices to guide neurosurgical procedures. Kollias et al (14) described their experience with an open-bore 0.5-T MR unit in 21 brain biopsies, and they found that this system enabled them to accurately localize and perform biopsy of brain lesions in all patients. Tronnier et al (15), Knauth et al (16), and Lewin et al (17) have used a 0.2-T interventional MR system for successful guidance in a variety of neurosurgical procedures. Liu et al (18) recently reported on the use of a 1.5-T interventional MR system to assist in the performance of neurosurgical procedures in more than 70 patients. The configuration of the 0.2-T and the 1.5-T units differed from that of the open-bore 0.5-T magnet used in the present study; although all interventional magnets have been shown to be effective in helping guide and evaluate neurosurgical procedures, the 0.2-T and 1.5-T systems required that the patient be moved out of the bore of the magnet during surgery. We review our preliminary experience with an open-bore 0.5-T magnet in intraoperative MR imaging guidance in the first 200 intracranial procedures performed at our institution.

	MATERIALS AND METHODS

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 
MR Imaging System
A prototypic open-configuration, superconducting, 0.5-T MR imager (Signa SP; GE Medical Systems, Milwaukee, Wis) was used. This unit allowed two operators to have simultaneous access to the patient through a 56-cm-wide vertical gap (Fig 1). The isocenter of the magnetic field was located midway between the two doughnuts of the open MR system. The field was homogeneous to within 10 ppm in a sphere with a 15-cm radius surrounding the isocenter. Outside of this sphere, homogeneity decreased rapidly with distance. The magnet was housed in a sterile procedure room in which all surgical and anesthesia equipment were fully MR compatible. A two-way audio system and a video camera mounted overhead in the procedure room allowed communication between the operators within the MR procedure room and the personnel at the console (11).

View larger version (94K): [in this window] [in a new window]   Figure 1a. Open-bore MR imaging system in the operating suite. (a) The 56-cm-wide gap in the magnet allows surgeons access to the patient. Two liquid crystal display screens (arrow) are positioned overhead to allow the radiologist to indicate the surgical approach to the surgeon. Plugs for the head lamps and the hand-held navigation device are located on the right half of the magnet. (b) View through the bore of the open-configuration MR imaging system shows the fixed portion of the Mayfield head frame (arrow) attached to the table. The MR-compatible anesthesia system is located to the right of the magnet.

View larger version (107K): [in this window] [in a new window]   Figure 1b. Open-bore MR imaging system in the operating suite. (a) The 56-cm-wide gap in the magnet allows surgeons access to the patient. Two liquid crystal display screens (arrow) are positioned overhead to allow the radiologist to indicate the surgical approach to the surgeon. Plugs for the head lamps and the hand-held navigation device are located on the right half of the magnet. (b) View through the bore of the open-configuration MR imaging system shows the fixed portion of the Mayfield head frame (arrow) attached to the table. The MR-compatible anesthesia system is located to the right of the magnet.

Receive and transmit surface coils were manufactured specifically for use with the intraoperative MR system, so that the coils could be wrapped around the patient's head to allow access to the intracranial lesion through a burr hole or craniotomy (8,19). Flexible double-loop surface coils provided radio-frequency excitation from both sides of the patient's head. Access to the patient was gained through a window in the loop or the space between the two loops. This flexibility allowed the positioning of patients in the supine, prone, or decubitus position to optimize the approach to the lesion.

Coils of two sizes were used: A larger coil (each loop: 22.5 x 23.5-cm inner diameter, 24.0 x 28.0-cm outer diameter) and a smaller coil (each loop: 15.0 x 11.0-cm inner diameter, 20.5 x 17.0-cm outer diameter). The larger coil provided deeper signal penetration and a larger area of coverage but had a decreased signal-to-noise ratio relative to that of the smaller coil. The smaller coil was used for smaller, more superficial lesions and provided a higher signal-to-noise ratio. A single-loop coil of similar size was used over the face for imaging during transsphenoidal resection. Intraoperatively acquired MR images were displayed at external consoles and on two 5-inch (diagonal width) liquid crystal display monitors mounted in the gap of the magnet.

Patients
Patients were referred to the intraoperative MR team for further evaluation of intracranial lesions visualized on previous imaging studies obtained with conventional imaging systems at our institution or elsewhere. In all cases, informed consent was obtained after the nature of the contemplated surgery and the role in surgery of the intraoperative MR system was fully explained to the patient.

Patients were considered to be ineligible for the procedure if they had an indwelling electromechanical device that could be affected by the intense magnetic fields of the MR device or if they had metallic clips in the brain that were not definitely known to be nonferromagnetic. Patients with a history of coronary disease were excluded because the magnet would induce distortions in the electrocardiogram, which would obscure any ischemia-related abnormalities such as ST segment elevations or T waves (20). Patients who were too large to fit into the bore of the magnet also were excluded. Any patient who had previously demonstrated a sensitivity to gadopentetate dimeglumine was considered to be ineligible unless his or her brain lesion was known, from previous imaging studies, not to show contrast agent enhancement.

Patients were aged 18–85 years (mean, 48.1 years). There were 97 men and 103 women. The majority of procedures in the intraoperative MR unit were performed with administration of general anesthesia. Procedures in 61 patients (33 biopsies, 28 craniotomies) were performed with the use of conscious sedation; these patients had lesions located near or in the "eloquent" brain areas, that is, regions that necessitated evaluation for motor and/or speech function at surgery during monitored, awake anesthesia.

Imaging Protocols and Procedures
Nonmagnetic materials were used for neurosurgical tools; these tools included 300–American Standard for Instrumentation (ASI)–grade stainless steel, titanium, brass, ceramics, and aluminum. Whereas titanium and ASI-300 stainless steel produced substantial artifact, the brass, ceramics, and carbon-fiber tools caused minimal artifact, and, therefore, imaging could be performed while these materials were in the field. The head holder (Ohio Medical, Cincinnati) was composed of carbon fiber and, therefore, was not associated with any distortion in the images. Ultrasonic scalpels (Ethicon, Cincinnati Ohio) or selector aspirators (Elekta, Atlanta, Ga) were used in the magnet for separating abnormal tissue from normal tissue on the basis of differences in consistency and cohesive forces. High-speed pneumatic drills (Midas Rex, Fort Worth, Tex) were similar to those used in the conventional operating room setting except that ball bearings were composed of ceramic instead of stainless steel. The MR-compatible bipolar electrocautery system (Codman and Shurtleff, Randolph, Mass) also was similar to that used in the conventional setting. Other devices such as intraoperative microscopes (Möller Microsurgical, Waldwick, NJ, and Studer Medical Engineering, Rhenfaal, Switzerland); headsets and other light sources (Cogent Light Technologies, Santa Clarita, Calif); transsphenoidal resection instruments (Aesculap, San Francisco, Calif); electrocardiographic, oxygen saturation, capnographic, and blood pressure monitors (MR Equipment, Bay Shore NY, and Bruker Instruments, Wissembourg, France); the anesthesia machine (Ohmeda, Madison, Wis); the ventilator (Omnivent, Topeka, Kan); and the peripheral nerve stimulators (Greatbach Scientific, Clarence, NY) were modified for use in the magnet (20–23).

All instruments were tested according to guidelines established by the manufacturer of the MR unit (MR safety and MR compatibility: test guidelines for Signa SP; GE Medical Systems) to ensure that the instruments would not be affected by the static magnetic field, magnetic field gradients, or radio-frequency pulses and to ensure that they would not cause noise or artifact on the MR images (24,25).

Instruments termed "MR safe" exhibited no torque or attraction to the magnet. The microscope and the anesthesia machine and monitors were constructed entirely from nonferromagnetic material. If the instrument was to be left in situ during scanning, it had to be constructed of a material that yielded no or little image artifact. Instrumentation and cables had to be properly shielded so that radio-frequency noise would not be received by the coil and manifested on the images.

The adapted bipolar and aspirator units were placed in a radio-frequency–tight enclosure (GE Medical Systems), which was bolted to the floor inside the magnet room. Custom-made long cables allowed the cabinet to be placed outside the 200-G (20-mT) line, beyond which most instruments functioned normally outside the field. The transformers for the anesthesia monitors were placed in an electrical room adjacent to the imager room; to remove radio-frequency noise, the direct-current power cables were filtered through connectors in the penetration panel before they were connected to the monitors. Fiberoptic or carbon-fiber electrocardiographic leads were used to eliminate the risk of burns to the patient (26). Pulse oximeter monitors also were equipped with fiberoptic leads. Components of the anesthesia machine and ventilator were powered by means of batteries designed for use in the MR environment, as were flashlights, laryngoscope handles, and the peripheral nerve stimulator. Nonconductive lines, such as the fiberoptic cables for the light sources, which were housed in the room adjacent to the imager room, were channeled through wave guides in the penetration panel without increasing the radio-frequency noise level in the imager room. The microscope was housed on a pneumatically powered articulating frame. Although the camera on the microscope functioned in the magnetic field, the camera controller unit was placed beyond the 20-mT line.

Interactive image update was accomplished by using a hand-held optical tracking device (Flashpoint; Image Guided Technologies, Boulder, Colo) mounted with three light-emitting diodes (Fig 2); three high-resolution infrared-sensitive video cameras located above the isocenter of the magnet were used to detect the emissions of the light-emitting diodes. We used specialized image-guidance software (REALTIME CONTROL; GE Medical Systems) implemented at an interactive workstation (model 4/670; Sun Microsystems, Mountain View, Calif) (8,9).

View larger version (108K): [in this window] [in a new window]   Figure 2. Hand-held optical tracking device. Note the three light-emitting diodes located at the ends of the arms of the device; these are used to establish both the plane of imaging and the proposed trajectory of a needle placed through the center (arrow) of the device. The cable consists of a shielded copper wire surrounded by a plastic covering.

For the biopsies, a titanium, 15-cm-long, 18- or 20-gauge needle (E-Z-Em, Westbury, NY) that produces minimal artifact and a tracking device were fixed to the flexible arm (J & J Professional and Codman, Raynham, Mass) of an MR-compatible head frame (OMI Surgical Products, Cincinnati, Ohio), which allowed precise control of the needle as it was advanced during near real-time imaging. If preprocedural imaging showed the brain lesion to be enhancing, imaging in the open-configuration magnet was performed after injection of gadopentetate dimeglumine (Magnevist; Berlex, Wayne, NJ). T1-weighted fast spin-echo MR images (repetition time msec/effective echo time msec, 400/17; one signal acquired; 22-cm field of view) were acquired every 14 seconds. Biopsies and resections of nonenhancing lesions were performed under guidance with a T2-weighted fast spin-echo technique (2,000/102 [effective], one signal acquired, 22-cm field of view), with which images were acquired every 20 seconds. Recent developments now allow T2-weighted imaging to be updated every 2 seconds by using a single-shot fast spin-echo technique (1,370/180 [effective], one-half signal acquired, 22-cm field of view, one-half–phase field of view); the single-shot fast spin-echo technique implements a half k-space acquisition to limit the specific absorption rate and to acquire the signal before T2 decay results in an unusable signal-to-noise ratio.

For the resections, the approach to the abnormality was determined by using the interactive near real-time system (as a virtual pointer or with interactive imaging-plane selection, as already described for the biopsies) or serial volume imaging. Either T1-weighted spin-echo (repetition time msec/echo time msec, 600/29; two signals acquired; 22-cm field of view) or three-dimensional fast spoiled gradient-echo (echo time, minimum full; 30° flip angle, 22-cm field of view) imaging after injection of gadopentetate dimeglumine or T2-weighted fast spin-echo imaging (4,000/85 [effective], one signal acquired, 22-cm field of view) was performed. Serial volume imaging provided updates of the surgical volume as frequently as was deemed necessary.

To localize viable tumor tissue within a region of brain tissue treated with high-dose radiation therapy, we performed dynamic imaging after injection of 20 mL of gadopentetate dimeglumine by using two-dimensional fast spoiled gradient-echo imaging (minimum echo time; 45° flip angle; one signal acquired; sequential, multiphase, interleaved, variable bandwidth and extended dynamic range, 10 phases per location; 22-cm field of view; 256 x 128 matrix; acquisition time, 1.87 seconds per image). This technique produced a series of images that depicted regional rates of enhancement through the tumor volume; previous data (12) have shown that this technique is highly accurate for the discrimination of recurrent tumor from radiation gliosis.

Heme-susceptibility gradient-recalled-echo imaging (600/9, 60; 30° flip angle) was used to determine the presence of acute hemorrhage. To assess for the presence of acute hemorrhage during and after the procedure, images through the surgical cavity were obtained through the resection bed by using short echo time (9 msec) and long echo time (60 msec) gradient-recalled-echo images. T2-weighted images were obtained as well. Foci of hypointense signal, which "bloomed" on the long echo time image but were not apparent on the T2-weighted image, were considered to represent deoxyhemoglobin in an area of acute hemorrhage. If such a collection was noted to exert mass effect on surrounding brain or to be rapidly increasing in size, it was removed under MR imaging guidance.

Intracranial cysts were examined by using serial T1-weighted images (600/29, two signals acquired, 256 x 128 matrix) obtained before and after approximately 1 mL of a solution of 0.1 mL of 0.5 mol/L gadopentetate dimeglumine, diluted in 0.9 mL of a normal saline solution, was injected into the cystic collection or surrounding cerebrospinal fluid (CSF) space. The patients provided separate informed consent for the use of a gadolinium-based contrast agent. If the collection communicated rapidly with the CSF spaces, no further intervention was performed. If no communication could be shown, the cyst was drained through a catheter or was opened after craniotomy (13).

Transsphenoidal resections were performed with the receive and transmit surface coil wrapped around the patient's head to allow access to the nose. A nasal speculum composed of titanium (Aesculap) was used for the procedure. Serial imaging was performed by using either T1-weighted (500/20, two signals acquired) or fast spoiled gradient-echo T1-weighted (variable repetition time, minimum full echo time; 30° flip angle) imaging in the sagittal and coronal planes. Dynamic imaging and heme-susceptibility sequences were used as needed.

	RESULTS

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 
From September 26, 1995, through October 7, 1997, 200 intracranial procedures were performed by using intraoperative MR imaging. These included 68 biopsies (60 for primary brain tumors, four for metastatic disease, three for abscesses, and one for demyelinating plaque), 111 craniotomies (95 for primary glial tumors, with radiation seeds placed in three patients; seven for metastatic tumors; six for vascular lesions, including four arteriovenous malformation resections and two cavernous angiomas; and three for meningiomas), 12 intracranial cyst evaluations (including two drainages), four subdural drainages, and five transsphenoidal pituitary gland resections. The results of interactive MR imaging guidance were satisfactory in all instances. In each case, the surgeon was accurately guided to the intracranial lesion, the progress of the procedure was assessed, and any immediate postoperative complications were evaluated before the incision or craniotomy was closed.

A similar protocol was followed in all cases. General anesthesia or intravenous sedation was administered in the preparation room outside the MR suite, and the patient was transferred to an MR-compatible operating table, which was then brought into the MR suite. The patients were moved into the bore of the magnet for the preprocedural setup, which included positioning of the coil on the basis of preoperative MR images (usually obtained with a high-field-strength [1.5-T] MR system), placement of the pins and clamp, and preparation of the surgical site. The site of the craniotomy was determined by the radiologist on the basis of T2-weighted images obtained with the interventional MR system, and the skull and dura were opened by using standard neurosurgical techniques.

The neurosurgical procedure was performed under MR guidance, after which the dura was closed, the bone flap was replaced, and the skin was closed, also according to standard protocol. The radiologist was present for the duration of the procedure, to interpret the images as they were obtained. The hand-held navigational device provided rapid imaging of a limited view of the brain relative to that provided by volumetric imaging. Therefore, interactive imaging was used only when rapid updating of information was required, such as during a biopsy, to direct a needle to an intracranial lesion. For resections, interactive imaging with the hand-held device was generally limited to demarcating foci of early dynamic enhancement in a lesion that was suggestive of recurrent tumor.

For the biopsies, the approach to each lesion was determined by using near real-time imaging. This allowed the operator to avoid important vascular structures and critical brain regions as the needle was advanced through the brain. Imaging in all planes allowed clinicians to be certain of the location of the biopsy needle when specimens were obtained, which thus obviated multiple needle passes and frozen-tissue sectioning (Fig 3). The cable connecting the optical tracking device consisted of a shielded copper wire, which, in turn, was grounded to the steel enclosure of the MR imager room. This shielding minimized radio-frequency noise emitted from the cable.

View larger version (139K): [in this window] [in a new window]   Figure 3a. T1-weighted MR images (400/17, one signal acquired) obtained during biopsy of an intracranial mass in a 42-year-old immunosuppressed man with left-sided weakness show an irregular 2.5-cm-diameter ring-enhancing lesion adjacent to the right frontal horn. The lesion was localized by using the interactive hand-held navigational system. (a) Axial image shows the projected biopsy site (x). (b, c) Serial sagittal images show the planned trajectory (dashed line) of the needle and the needle itself (arrow) as it is gradually advanced (b) toward and (c) into the lesion. (d) Axial and (e) coronal MR images obtained after the biopsy show air (arrow) in the lesion (x), which verifies the accuracy of the biopsy. Histopathologic analysis demonstrated non-Hodgkin lymphoma. The dashed line in b–d indicates the needle trajectory; the junction between the long and short dashes indicates the location of "virtual" needle tip.

View larger version (121K): [in this window] [in a new window]   Figure 3b. T1-weighted MR images (400/17, one signal acquired) obtained during biopsy of an intracranial mass in a 42-year-old immunosuppressed man with left-sided weakness show an irregular 2.5-cm-diameter ring-enhancing lesion adjacent to the right frontal horn. The lesion was localized by using the interactive hand-held navigational system. (a) Axial image shows the projected biopsy site (x). (b, c) Serial sagittal images show the planned trajectory (dashed line) of the needle and the needle itself (arrow) as it is gradually advanced (b) toward and (c) into the lesion. (d) Axial and (e) coronal MR images obtained after the biopsy show air (arrow) in the lesion (x), which verifies the accuracy of the biopsy. Histopathologic analysis demonstrated non-Hodgkin lymphoma. The dashed line in b–d indicates the needle trajectory; the junction between the long and short dashes indicates the location of "virtual" needle tip.

View larger version (120K): [in this window] [in a new window]   Figure 3c. T1-weighted MR images (400/17, one signal acquired) obtained during biopsy of an intracranial mass in a 42-year-old immunosuppressed man with left-sided weakness show an irregular 2.5-cm-diameter ring-enhancing lesion adjacent to the right frontal horn. The lesion was localized by using the interactive hand-held navigational system. (a) Axial image shows the projected biopsy site (x). (b, c) Serial sagittal images show the planned trajectory (dashed line) of the needle and the needle itself (arrow) as it is gradually advanced (b) toward and (c) into the lesion. (d) Axial and (e) coronal MR images obtained after the biopsy show air (arrow) in the lesion (x), which verifies the accuracy of the biopsy. Histopathologic analysis demonstrated non-Hodgkin lymphoma. The dashed line in b–d indicates the needle trajectory; the junction between the long and short dashes indicates the location of "virtual" needle tip.

View larger version (131K): [in this window] [in a new window]   Figure 3d. T1-weighted MR images (400/17, one signal acquired) obtained during biopsy of an intracranial mass in a 42-year-old immunosuppressed man with left-sided weakness show an irregular 2.5-cm-diameter ring-enhancing lesion adjacent to the right frontal horn. The lesion was localized by using the interactive hand-held navigational system. (a) Axial image shows the projected biopsy site (x). (b, c) Serial sagittal images show the planned trajectory (dashed line) of the needle and the needle itself (arrow) as it is gradually advanced (b) toward and (c) into the lesion. (d) Axial and (e) coronal MR images obtained after the biopsy show air (arrow) in the lesion (x), which verifies the accuracy of the biopsy. Histopathologic analysis demonstrated non-Hodgkin lymphoma. The dashed line in b–d indicates the needle trajectory; the junction between the long and short dashes indicates the location of "virtual" needle tip.

View larger version (116K): [in this window] [in a new window]   Figure 3e. T1-weighted MR images (400/17, one signal acquired) obtained during biopsy of an intracranial mass in a 42-year-old immunosuppressed man with left-sided weakness show an irregular 2.5-cm-diameter ring-enhancing lesion adjacent to the right frontal horn. The lesion was localized by using the interactive hand-held navigational system. (a) Axial image shows the projected biopsy site (x). (b, c) Serial sagittal images show the planned trajectory (dashed line) of the needle and the needle itself (arrow) as it is gradually advanced (b) toward and (c) into the lesion. (d) Axial and (e) coronal MR images obtained after the biopsy show air (arrow) in the lesion (x), which verifies the accuracy of the biopsy. Histopathologic analysis demonstrated non-Hodgkin lymphoma. The dashed line in b–d indicates the needle trajectory; the junction between the long and short dashes indicates the location of "virtual" needle tip.

View larger version (137K): [in this window] [in a new window]   Figure 4a. T2-weighted (4,000/85, two signals acquired) MR imaging–guided resection of a low-grade astrocytoma in a 40-year-old man with recent onset of seizures. (a) Coronal MR image shows a well-defined high-signal-intensity abnormality (large arrow) in the left frontal lobe; a capsule containing cod-liver oil (small arrow) has been placed on the scalp to demarcate the inferior border of the lesion, in preparation for craniotomy. (b) Axial MR image shows a well-defined high-signal-intensity abnormality (large arrow) in the left frontal lobe; a capsule containing cod-liver oil (small arrow) has been placed on the scalp to demarcate the posterior border of the lesion, in preparation for craniotomy. (c) Axial MR image shows the site of craniotomy (arrows), which was performed under MR imaging guidance. (d) Axial MR image shows that after the dura was removed, abnormal tissue (arrow) herniated through the craniotomy, distorting the preoperative anatomy. MR imaging allowed the radiologist and surgeons to compensate for this anatomic change. (e) Axial MR image shows that after the bulk of the lesion was removed, a small focus of residual tumor, not readily visible to the surgeon, was demarcated with a tuberculin syringe filled with saline solution (long arrow), which acted as a high-signal-intensity pointer. Owing to the proximity of the lesion to the motor strip (short arrow), the tumor was not resected.

View larger version (101K): [in this window] [in a new window]   Figure 4b. T2-weighted (4,000/85, two signals acquired) MR imaging–guided resection of a low-grade astrocytoma in a 40-year-old man with recent onset of seizures. (a) Coronal MR image shows a well-defined high-signal-intensity abnormality (large arrow) in the left frontal lobe; a capsule containing cod-liver oil (small arrow) has been placed on the scalp to demarcate the inferior border of the lesion, in preparation for craniotomy. (b) Axial MR image shows a well-defined high-signal-intensity abnormality (large arrow) in the left frontal lobe; a capsule containing cod-liver oil (small arrow) has been placed on the scalp to demarcate the posterior border of the lesion, in preparation for craniotomy. (c) Axial MR image shows the site of craniotomy (arrows), which was performed under MR imaging guidance. (d) Axial MR image shows that after the dura was removed, abnormal tissue (arrow) herniated through the craniotomy, distorting the preoperative anatomy. MR imaging allowed the radiologist and surgeons to compensate for this anatomic change. (e) Axial MR image shows that after the bulk of the lesion was removed, a small focus of residual tumor, not readily visible to the surgeon, was demarcated with a tuberculin syringe filled with saline solution (long arrow), which acted as a high-signal-intensity pointer. Owing to the proximity of the lesion to the motor strip (short arrow), the tumor was not resected.

View larger version (109K): [in this window] [in a new window]   Figure 4c. T2-weighted (4,000/85, two signals acquired) MR imaging–guided resection of a low-grade astrocytoma in a 40-year-old man with recent onset of seizures. (a) Coronal MR image shows a well-defined high-signal-intensity abnormality (large arrow) in the left frontal lobe; a capsule containing cod-liver oil (small arrow) has been placed on the scalp to demarcate the inferior border of the lesion, in preparation for craniotomy. (b) Axial MR image shows a well-defined high-signal-intensity abnormality (large arrow) in the left frontal lobe; a capsule containing cod-liver oil (small arrow) has been placed on the scalp to demarcate the posterior border of the lesion, in preparation for craniotomy. (c) Axial MR image shows the site of craniotomy (arrows), which was performed under MR imaging guidance. (d) Axial MR image shows that after the dura was removed, abnormal tissue (arrow) herniated through the craniotomy, distorting the preoperative anatomy. MR imaging allowed the radiologist and surgeons to compensate for this anatomic change. (e) Axial MR image shows that after the bulk of the lesion was removed, a small focus of residual tumor, not readily visible to the surgeon, was demarcated with a tuberculin syringe filled with saline solution (long arrow), which acted as a high-signal-intensity pointer. Owing to the proximity of the lesion to the motor strip (short arrow), the tumor was not resected.

View larger version (91K): [in this window] [in a new window]   Figure 4d. T2-weighted (4,000/85, two signals acquired) MR imaging–guided resection of a low-grade astrocytoma in a 40-year-old man with recent onset of seizures. (a) Coronal MR image shows a well-defined high-signal-intensity abnormality (large arrow) in the left frontal lobe; a capsule containing cod-liver oil (small arrow) has been placed on the scalp to demarcate the inferior border of the lesion, in preparation for craniotomy. (b) Axial MR image shows a well-defined high-signal-intensity abnormality (large arrow) in the left frontal lobe; a capsule containing cod-liver oil (small arrow) has been placed on the scalp to demarcate the posterior border of the lesion, in preparation for craniotomy. (c) Axial MR image shows the site of craniotomy (arrows), which was performed under MR imaging guidance. (d) Axial MR image shows that after the dura was removed, abnormal tissue (arrow) herniated through the craniotomy, distorting the preoperative anatomy. MR imaging allowed the radiologist and surgeons to compensate for this anatomic change. (e) Axial MR image shows that after the bulk of the lesion was removed, a small focus of residual tumor, not readily visible to the surgeon, was demarcated with a tuberculin syringe filled with saline solution (long arrow), which acted as a high-signal-intensity pointer. Owing to the proximity of the lesion to the motor strip (short arrow), the tumor was not resected.

View larger version (105K): [in this window] [in a new window]   Figure 4e. T2-weighted (4,000/85, two signals acquired) MR imaging–guided resection of a low-grade astrocytoma in a 40-year-old man with recent onset of seizures. (a) Coronal MR image shows a well-defined high-signal-intensity abnormality (large arrow) in the left frontal lobe; a capsule containing cod-liver oil (small arrow) has been placed on the scalp to demarcate the inferior border of the lesion, in preparation for craniotomy. (b) Axial MR image shows a well-defined high-signal-intensity abnormality (large arrow) in the left frontal lobe; a capsule containing cod-liver oil (small arrow) has been placed on the scalp to demarcate the posterior border of the lesion, in preparation for craniotomy. (c) Axial MR image shows the site of craniotomy (arrows), which was performed under MR imaging guidance. (d) Axial MR image shows that after the dura was removed, abnormal tissue (arrow) herniated through the craniotomy, distorting the preoperative anatomy. MR imaging allowed the radiologist and surgeons to compensate for this anatomic change. (e) Axial MR image shows that after the bulk of the lesion was removed, a small focus of residual tumor, not readily visible to the surgeon, was demarcated with a tuberculin syringe filled with saline solution (long arrow), which acted as a high-signal-intensity pointer. Owing to the proximity of the lesion to the motor strip (short arrow), the tumor was not resected.

View larger version (45K): [in this window] [in a new window]   Figure 5a. Dynamic contrast-enhanced MR imaging–guided resection of possible recurrence of intracranial metastatic disease in a 64-year-old woman. The patient had a recent onset of headaches and progressive visual loss in the left visual field 15 months after undergoing high-dose radiation therapy for an intracranial metastasis from lung cancer. MR imaging demonstrated an irregularly shaped enhancing lesion in the left occipital lobe, which represented a combination of recurrent tumor and gliosis due to radiation changes. The patient is prone. (a) Axial two-dimensional fast spoiled gradient-echo MR image (minimum echo time; flip angle, 45°; section thickness, 5 mm; one signal acquired; acquisition time, 1.87 seconds per image) obtained immediately after bolus injection of 20 mL of 0.5 mol/L gadopentetate dimeglumine. A pixel-by-pixel curve fit was performed, and the data were color coded according to the parameters of the fit. This color map has been superimposed on an axial T1-weighted image (600/29, two signals acquired). Foci of early enhancement are shown in magenta, which indicates the likely presence of recurrent tumor. (b) Axial T1-weighted image (600/29, two signals acquired) obtained, with the patient in the prone position, after partial resection of the treated tumor bed. A tuberculin syringe (arrow) filled with the patient's gadolinium-enhanced blood acts as a high-signal-intensity pointer and indicates a focus of early enhancement in the treated tumor bed. Histopathologic analysis demonstrated recurrent metastatic tumor.

View larger version (67K): [in this window] [in a new window]   Figure 5b. Dynamic contrast-enhanced MR imaging–guided resection of possible recurrence of intracranial metastatic disease in a 64-year-old woman. The patient had a recent onset of headaches and progressive visual loss in the left visual field 15 months after undergoing high-dose radiation therapy for an intracranial metastasis from lung cancer. MR imaging demonstrated an irregularly shaped enhancing lesion in the left occipital lobe, which represented a combination of recurrent tumor and gliosis due to radiation changes. The patient is prone. (a) Axial two-dimensional fast spoiled gradient-echo MR image (minimum echo time; flip angle, 45°; section thickness, 5 mm; one signal acquired; acquisition time, 1.87 seconds per image) obtained immediately after bolus injection of 20 mL of 0.5 mol/L gadopentetate dimeglumine. A pixel-by-pixel curve fit was performed, and the data were color coded according to the parameters of the fit. This color map has been superimposed on an axial T1-weighted image (600/29, two signals acquired). Foci of early enhancement are shown in magenta, which indicates the likely presence of recurrent tumor. (b) Axial T1-weighted image (600/29, two signals acquired) obtained, with the patient in the prone position, after partial resection of the treated tumor bed. A tuberculin syringe (arrow) filled with the patient's gadolinium-enhanced blood acts as a high-signal-intensity pointer and indicates a focus of early enhancement in the treated tumor bed. Histopathologic analysis demonstrated recurrent metastatic tumor.

View larger version (144K): [in this window] [in a new window]   Figure 6a. Axial T1-weighted intraoperative MR images (600/29, two signals acquired) of resection of treated glioblastoma multiforme in a 54-year-old woman show the appearance of various artifacts. (a) Image obtained after a craniotomy was performed shows a heterogeneously enhancing mass in the right frontal lobe. The tip of the surgeon's gloved finger (arrow) demarcates the medial portion of the tumor bed. (b) During the course of the resection, air in the subdural space has caused brain to shift away from the skull and compress the right lateral ventricle (arrow). With conventional stereotactic methods, this intraoperative shift of brain structures cannot be predicted. (c) Image obtained through the inferior portion of the tumor bed shows a cotton ball (arrow), which was placed in the resection cavity to absorb blood. (d) Image obtained after the enhancing tissue was removed (histopathologic analysis demonstrated a combination of gliosis and residual tumor) shows iodine 125 radiation seeds (arrows) that were implanted under MR imaging guidance.

View larger version (144K): [in this window] [in a new window]   Figure 6b. Axial T1-weighted intraoperative MR images (600/29, two signals acquired) of resection of treated glioblastoma multiforme in a 54-year-old woman show the appearance of various artifacts. (a) Image obtained after a craniotomy was performed shows a heterogeneously enhancing mass in the right frontal lobe. The tip of the surgeon's gloved finger (arrow) demarcates the medial portion of the tumor bed. (b) During the course of the resection, air in the subdural space has caused brain to shift away from the skull and compress the right lateral ventricle (arrow). With conventional stereotactic methods, this intraoperative shift of brain structures cannot be predicted. (c) Image obtained through the inferior portion of the tumor bed shows a cotton ball (arrow), which was placed in the resection cavity to absorb blood. (d) Image obtained after the enhancing tissue was removed (histopathologic analysis demonstrated a combination of gliosis and residual tumor) shows iodine 125 radiation seeds (arrows) that were implanted under MR imaging guidance.

View larger version (147K): [in this window] [in a new window]   Figure 6c. Axial T1-weighted intraoperative MR images (600/29, two signals acquired) of resection of treated glioblastoma multiforme in a 54-year-old woman show the appearance of various artifacts. (a) Image obtained after a craniotomy was performed shows a heterogeneously enhancing mass in the right frontal lobe. The tip of the surgeon's gloved finger (arrow) demarcates the medial portion of the tumor bed. (b) During the course of the resection, air in the subdural space has caused brain to shift away from the skull and compress the right lateral ventricle (arrow). With conventional stereotactic methods, this intraoperative shift of brain structures cannot be predicted. (c) Image obtained through the inferior portion of the tumor bed shows a cotton ball (arrow), which was placed in the resection cavity to absorb blood. (d) Image obtained after the enhancing tissue was removed (histopathologic analysis demonstrated a combination of gliosis and residual tumor) shows iodine 125 radiation seeds (arrows) that were implanted under MR imaging guidance.

View larger version (133K): [in this window] [in a new window]   Figure 6d. Axial T1-weighted intraoperative MR images (600/29, two signals acquired) of resection of treated glioblastoma multiforme in a 54-year-old woman show the appearance of various artifacts. (a) Image obtained after a craniotomy was performed shows a heterogeneously enhancing mass in the right frontal lobe. The tip of the surgeon's gloved finger (arrow) demarcates the medial portion of the tumor bed. (b) During the course of the resection, air in the subdural space has caused brain to shift away from the skull and compress the right lateral ventricle (arrow). With conventional stereotactic methods, this intraoperative shift of brain structures cannot be predicted. (c) Image obtained through the inferior portion of the tumor bed shows a cotton ball (arrow), which was placed in the resection cavity to absorb blood. (d) Image obtained after the enhancing tissue was removed (histopathologic analysis demonstrated a combination of gliosis and residual tumor) shows iodine 125 radiation seeds (arrows) that were implanted under MR imaging guidance.

View larger version (99K): [in this window] [in a new window]   Figure 7a. Sagittal T1-weighted MR images (600/29, two signals acquired) show an intracranial cyst in a 34-year-old man. The cyst was localized and drained under intraoperative MR imaging guidance. (a) A supracerebellar cystic collection (arrow) is visible. The surgeon's gloved finger is pointing toward the lesion, demarcating the planned trajectory of the needle for drainage. (b) After 0.1 mL of 0.5 mol/L gadopentetate dimeglumine (curved arrow) was injected into the cyst, no communication with the surrounding CSF was demonstrated. An intracranial catheter (straight arrows) was used to drain the cyst under MR imaging guidance.

View larger version (104K): [in this window] [in a new window]   Figure 7b. Sagittal T1-weighted MR images (600/29, two signals acquired) show an intracranial cyst in a 34-year-old man. The cyst was localized and drained under intraoperative MR imaging guidance. (a) A supracerebellar cystic collection (arrow) is visible. The surgeon's gloved finger is pointing toward the lesion, demarcating the planned trajectory of the needle for drainage. (b) After 0.1 mL of 0.5 mol/L gadopentetate dimeglumine (curved arrow) was injected into the cyst, no communication with the surrounding CSF was demonstrated. An intracranial catheter (straight arrows) was used to drain the cyst under MR imaging guidance.

View larger version (110K): [in this window] [in a new window]   Figure 8a. Sagittal T1-weighted (500/20, two signals acquired) MR imaging–guided resection of pituitary macroadenoma in a 42-year-old man. The patient is supine, with the head oriented in the right decubitus position. (a) Contrast-enhanced MR image shows a 2-cm enhancing mass in the sella turcica (white arrow). Note the metallic artifact from the titanium nasal speculum (black arrow). (b) MR image obtained after resection of the mass shows a sterile compressed sponge (arrow) packed into the surgical site.

View larger version (111K): [in this window] [in a new window]   Figure 8b. Sagittal T1-weighted (500/20, two signals acquired) MR imaging–guided resection of pituitary macroadenoma in a 42-year-old man. The patient is supine, with the head oriented in the right decubitus position. (a) Contrast-enhanced MR image shows a 2-cm enhancing mass in the sella turcica (white arrow). Note the metallic artifact from the titanium nasal speculum (black arrow). (b) MR image obtained after resection of the mass shows a sterile compressed sponge (arrow) packed into the surgical site.

Patients who underwent craniotomy were positioned in the magnet for an average of 6 hours 42 minutes; the surgery lasted an average of 5 hours 8 minutes, whereas a craniotomy performed in the conventional operating room averaged approximately 5 hours.

For both biopsies and craniotomies, the setup time in the magnet was approximately 1 hours; this included imaging prior to surgery to localize the lesion and determine the surgical approach. Although this portion of the procedure is unique to intraoperative MR, neurosurgeons who use conventional techniques often use stereotactic localization, which requires placement of a head frame, additional CT imaging, and encoding of computerized data. Thus, the time for surgery in the open magnet was comparable to that for procedures performed in the conventional operating room.

There were four complications noted in our patients: Two patients (1%) experienced a postoperative intracranial infection or inflammatory process, and two (1%) required repeat surgery for removal of an intracranial hematoma (1%). One patient developed an intracerebral infection after a craniotomy; the infection was treated with intravenous administration of antibiotics. Another patient developed a sterile meningitis after a gadolinium chelate was injected into the CSF; the meningitis was treated successfully with intravenous administration of steroids.

Although a small amount of hemorrhage in the resection margin was often present at the end of a procedure (Fig 9), mass effect due to these collections necessitated removal in two patients. After an intracranial biopsy was performed in one patient, a small hematoma was noted at the biopsy site, and the decision was made to perform a craniotomy to resect the clot. Another patient had undergone craniotomy for resection of a high-grade tumor, and an intraparenchymal hemorrhage was noted during routine postoperative heme-susceptibility MR imaging after the dura had been closed. The surgeon reopened the dura to remove the hemorrhage. During the 2 years of this study, the incidence of intracranial infection and postoperative hematoma that necessitated repeat surgery were each approximately 1% after conventional neurosurgery at our institution. Thus, the incidence of complications encountered in procedures performed with the open magnet was similar to that encountered in the conventional setting.

View larger version (128K): [in this window] [in a new window]   Figure 9a. Hyperacute hemorrhage in a 19-year-old woman after resection of a low-grade astrocytoma. (a) Axial T1-weighted (600/29, two signals acquired) nonenhanced MR image obtained immediately after surgery shows an area of heterogeneous and predominantly isointense signal in a surgical cavity (arrow). (b) Axial T2-weighted (4,000/85, two signals acquired) MR image obtained immediately after surgery shows an area of heterogeneous and predominantly isointense signal intensity in a surgical cavity (arrow). (c) Axial gradient-recalled-echo MR image (600/9, 30° flip angle) shows several small foci of low T2 signal intensity (arrow). (d) On the longer echo time gradient-recalled-echo MR image (600/60, 30° flip angle), these foci bloom because of a hyperacute blood collection (arrow) that was not evident on the conventional MR images.

View larger version (133K): [in this window] [in a new window]   Figure 9b. Hyperacute hemorrhage in a 19-year-old woman after resection of a low-grade astrocytoma. (a) Axial T1-weighted (600/29, two signals acquired) nonenhanced MR image obtained immediately after surgery shows an area of heterogeneous and predominantly isointense signal in a surgical cavity (arrow). (b) Axial T2-weighted (4,000/85, two signals acquired) MR image obtained immediately after surgery shows an area of heterogeneous and predominantly isointense signal intensity in a surgical cavity (arrow). (c) Axial gradient-recalled-echo MR image (600/9, 30° flip angle) shows several small foci of low T2 signal intensity (arrow). (d) On the longer echo time gradient-recalled-echo MR image (600/60, 30° flip angle), these foci bloom because of a hyperacute blood collection (arrow) that was not evident on the conventional MR images.

View larger version (98K): [in this window] [in a new window]   Figure 9c. Hyperacute hemorrhage in a 19-year-old woman after resection of a low-grade astrocytoma. (a) Axial T1-weighted (600/29, two signals acquired) nonenhanced MR image obtained immediately after surgery shows an area of heterogeneous and predominantly isointense signal in a surgical cavity (arrow). (b) Axial T2-weighted (4,000/85, two signals acquired) MR image obtained immediately after surgery shows an area of heterogeneous and predominantly isointense signal intensity in a surgical cavity (arrow). (c) Axial gradient-recalled-echo MR image (600/9, 30° flip angle) shows several small foci of low T2 signal intensity (arrow). (d) On the longer echo time gradient-recalled-echo MR image (600/60, 30° flip angle), these foci bloom because of a hyperacute blood collection (arrow) that was not evident on the conventional MR images.

View larger version (86K): [in this window] [in a new window]   Figure 9d. Hyperacute hemorrhage in a 19-year-old woman after resection of a low-grade astrocytoma. (a) Axial T1-weighted (600/29, two signals acquired) nonenhanced MR image obtained immediately after surgery shows an area of heterogeneous and predominantly isointense signal in a surgical cavity (arrow). (b) Axial T2-weighted (4,000/85, two signals acquired) MR image obtained immediately after surgery shows an area of heterogeneous and predominantly isointense signal intensity in a surgical cavity (arrow). (c) Axial gradient-recalled-echo MR image (600/9, 30° flip angle) shows several small foci of low T2 signal intensity (arrow). (d) On the longer echo time gradient-recalled-echo MR image (600/60, 30° flip angle), these foci bloom because of a hyperacute blood collection (arrow) that was not evident on the conventional MR images.

	DISCUSSION

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

Although the data obtained at stereotactic surgery are accurate and reproducible, these techniques have some serious limitations. One potential source of error at conventional neurosurgery is the choice of the location and size of the craniotomy, which usually are determined by the surgeon on the basis of preoperative imaging studies. Surgery may be made difficult because of an inauspicious choice of craniotomy, but intraoperative MR imaging allows the location and size of the craniotomy to be tailored to fit the precise location and contours of the lesion (Figs 3, 4).

When a biopsy or craniotomy is planned in which conventional stereotactic methods are to be used, it is necessarily assumed that the internal relationships of the brain will be maintained throughout the procedure, but this may not be the case. The normal brain has a relatively soft consistency and is surrounded by structures that contain CSF; therefore, instrumentation may alter intracranial anatomic relationships. Surgery may allow air to enter the extraaxial or intraventricular spaces; drainage of an intratumoral cyst or resection of a portion of a tumor may alter mass effect on surrounding brain; and edema may be induced in brain tissue by the act of resecting an adjacent lesion. Thus, preoperative information may become increasingly inaccurate as a procedure progresses (Fig 7). Continuous updating of the images by using intraoperative MR imaging permits the clinician to make allowances for such anatomic changes as they occur.

Instruments
The ultimate success of the intraoperative MR system is dependent on the availability of instruments that are designed for use in the MR environment, for both safety and imaging considerations (24,25). The instruments must be composed of nonferrous materials so that they will not be attracted by the magnetic field, and they must not produce excessive artifact on images obtained while the instruments are in the field.

These considerations become increasingly critical at medium and high field strengths. Such MR-safe instruments are continuously being developed for MR imaging. The most commonly used materials for use in intraoperative MR are 300-ASI–grade stainless steel or brass (both of which are relatively soft but are suitable for disposable instruments such as drill bits or forceps), titanium (used for many instruments, including the nasal speculum for transsphenoidal resections, needle holders, retractors, forceps, and scissors), ceramics (used for scalpel blades and scissors), and aluminum and nickel (fashioned into retractors and knife handles). Carbon fiber, which is used in the construction of retractors and the Mayfield head holder, is unaffected by the magnetic field; although the skull pins in the head holder are composed of titanium, their artifact does not extend beneath the cranium. It is important to note that the carbon fiber in the head frame does not produce spatial distortion, so imaging can be performed while these materials are in the field. Drills, selector aspirators, electrocautery systems, intraoperative microscopes, and light sources all have been adapted for use in intraoperative MR imaging, and it is expected that the full armamentarium of neurosurgical tools soon will be available in MR-compatible form.

Monitoring of the patient within the MR system poses some special challenges (11,20–25). Carbon-fiber or fiberoptic leads are routinely used to monitor the electrocardiogram during an MR procedure, but the leads must be close together to avoid looping of the wires, in which case an electric current may be induced by the magnetic field (26). The determination of cardiac ischemic changes likely will remain a problem, however, owing to MR-induced distortions of the electrocardiogram (19). In addition, cortical mapping with bipolar cortical stimulation is not yet available for the interventional MR system; thus, it may be relatively more difficult to remove lesions located near eloquent cortex than it is in the conventional operating room, where such techniques are routinely used. However, MR-compatible bipolar cortical stimulators, which are in development, and even functional MR imaging techniques may soon provide the information necessary to avoid sensitive cortical regions and white matter tracts in the intraoperative magnet.

Biopsies
Our experience with 68 biopsies indicates that MR imaging provides the accuracy necessary for localization and safe targeting. Even more important, MR imaging can be used to alter the trajectory as needed during a procedure. Continuous MR imaging with the hand-held optical tracking system allowed the operator to maintain a "fix" on an intracranial lesion as a biopsy needle was advanced toward it, and three-dimensional imaging allowed the surgeon to avoid sensitive intracranial structures. The accuracy of approach can be gauged in three planes, so it may not be necessary to await results from frozen biopsy specimens to verify whether the lesion has been sampled (Fig 3). Our experience is similar to that of Kollias et al (14) in 21 patients who underwent biopsy; they reported that satisfactory tissue was obtained without complications in all patients, with an imaging time of approximately 1 hour per patient.

Craniotomies
From the standpoint of the neurosurgeon, perhaps the most important aspect of intraoperative MR imaging is the ability to visualize the MR image–delineated boundaries of a tumor during surgery. Nevertheless, there are other important advantages of MR imaging–guided surgery. At conventional surgery, the surgeon's view of the brain surface through the craniotomy is restricted, and deeper structures can be visualized only after the more superficial tissue has been removed. Intraoperative MR imaging provides an accurate depiction of the entire surgical volume during the procedure and, even more important, allows updating of the changing anatomic relationships as the procedure progresses. This near real-time guidance can limit the damage to functionally intact, normal brain tissue. The open-bore magnet used in our series provided immediate imaging feedback to the surgeon, in contrast to MR systems (15–18) that require the patient to be withdrawn from the magnet for surgery and advanced back into the magnet for imaging; some of those systems, however, are 1.5-T imagers that can provide images with better resolution and can be used with more rapid imaging sequences than those that are possible with an open-bore 0.5-T magnet (18).

The margins of brain lesions usually can be delineated more accurately with MR imaging than with direct visual inspection. Low-grade astrocytomas are nonenhancing lesions that generally have well-defined margins on T1- and T2-weighted images. Nevertheless, these tumors are difficult to distinguish intraoperatively from normal brain; thus, there is a tendency for surgeons to be conservative and to perform incomplete tumor resections, especially in the vicinity of eloquent cortex. Nicolato et al (27) showed that the survival of patients with low-grade astrocytoma was strongly correlated with the extent of tumor resection, because chemotherapy and radiation therapy are not of proven effectiveness in the control of these tumors.

High-grade gliomas (anaplastic astrocytomas and glioblastomas) tend to enhance heterogeneously, indicating a breakdown of the blood-brain barrier. This is a useful characteristic that generally indicates the presence of tumor in untreated cases. However, these lesions tend to infiltrate brain tissue without evident enhancement, and the margins of such a tumor may be difficult to assess visually or with imaging methods. Peritumoral edema may be extensive, but it generally is not possible to determine whether the edema is reactive or the edematous tissue contains malignant tumor cells. Although there have been conflicting reports about the value of extensive cytoreductive surgery in high-grade gliomas, Albert et al (28) indicated that in patients who have undergone resection of a high-grade glioma, the presence of residual enhancing tissue was correlated with decreased patient survival. We therefore generally attempt to delimit the enhancing perimeter of a high-grade intracranial tumor and to direct the surge