Custom Intraoperative Magnet Designs

The first iMRI system was developed through a collaborative effort between physicians at Brigham and Women’s Hospital in Boston and engineers at General Electric Medical Systems. The design for the new magnet was paramount in the evolution of iMRI because it strayed from the traditional closed-coil configurations that prohibited patient access during scanning.⁹ Two superconducting magnets with bores 60 cm in diameter were vertically oriented in parallel, creating a 56-cm-wide gap between the magnets that provided nearly full access to the patient on either side of the OR table. The Signa SP design (General Electric Medical Systems, Milwaukee, WI), more commonly known as the “double doughnut,” positioned the patient’s head within the 0.5-T imaging isocenter located both between the two magnets and between the two surgical access sites, allowing simultaneous imaging and surgery without the need to move the patient (Fig. 120-1). The scanner also had a three-dimensional navigational system (Flashpoint, Integrated Technologies, Boulder, CO) that provided tracking capabilities within the imaging area. Because surgery took place within the magnetic field, MRI-compatible surgical and anesthesia equipment was required. Also of note was a specially constructed MRI-compatible microscope that provided the magnification necessary for neurosurgery but did not have many of the features offered on other neurosurgical microscopes, such as autofocus or autozoom.³ As the first of its kind, this iMRI system offered the unprecedented ability of producing near real-time intraoperative magnetic resonance images without the need to move the patient or magnet, while still providing adequate surgical access. The disadvantages included lower image quality relative to systems with high-field magnets, the need for more costly magnetic resonance–compatible equipment, and the somewhat limited surgical maneuverability given the confines of the magnets.¹

FIGURE 120-1 The General Electric “double doughnut.” Two vertically oriented magnets create a gap on either side of the operating table through which surgeons can access the patient. The construction allows for near real-time image acquisition and eliminates the need for patient or magnet movement between operating and scanning. The 0.5-T field strength, however, necessitates the use of custom magnetic resonance imaging–compatible surgical instruments and equipment.

(Image courtesy of Dr. Ferenc A. Jolesz, Department of Radiology, Brigham & Women’s Hospital, Harvard Medical School, Boston, MA.)

After the debut of the Signa SP into the neurosurgical world, other iMRI systems emerged that used novel magnet designs to optimize patient access with intraoperative imaging. The PoleStar magnet (Odin Medical Technologies, Yokneam, Israel) was one such design, which consisted of two vertical parallel disks containing ceramic magnets placed 25 cm apart and connected by a U-shaped arm that wrapped around the patient’s head from below (Fig. 120-2). The portability of the magnet limited its field intensity (0.12 to 0.15 T) but granted it great compatibility within the traditional OR. Given its low field strength, traditional surgical tools and equipment could be used and, if needed, the magnet could be placed inside its iron storage cabinet to virtually eliminate the magnetic field within the OR. Its small size allowed it to be wheeled to and placed beneath the patient’s head when needed for scanning, without any additional floor reinforcement. An optional radiofrequency (RF) shield that extended around the patient and magnet like a tent during imaging was available for ORs that were not RF shielded. Moreover, the surgical navigation system and the scanner could be operated by the neurosurgeon or OR staff, obviating the need for a neuroradiologist or a specific MRI technician.^3,13

FIGURE 120-2 The Medtronic PoleStar. This low-field intraoperative magnetic resonance imaging (iMRI) system consists of two vertically oriented, parallel magnets constructed as a U-shaped arm to fit underneath the head of any standard operating room table. The magnet’s small size and field strength allow it to be rolled in and out of a radiofrequency-shielded storage unit within the operating room and, in addition, permit the use of standard MRI-incompatible instruments and equipment.

(Image courtesy of Medtronic Navigation, Louisville, CO.)

The physicians and scientists from Toronto Western Hospital developed a vertical gap system that took the first step toward a mobile patient model in the evolution of iMRI. This system consisted of a traditional 0.2-T horizontal gap permanent magnet system that was rotated 90 degrees to create a vertical gap. While the patient was positioned within the magnet’s isocenter, simple procedures such as catheter and shunt insertions could be performed by the single surgeon who stood within the vertical gap. By sliding the patient, head first, 1 to 1.5 meters away from the scanning position, a surgical position with 270 degrees of access could be achieved. Throughout the movement of the patient, overhead cameras maintained patient registration, allowing shifts between surgery and scanning without the need for reregistering anatomic coordinates to magnetic resonance images.³ Despite its decommissioning in 2003, the 5-year trial with this vertical gap system demonstrated that moving the patient in and out of the scanner was not the serious hindrance it was once believed to be. From this experience, others began to explore the possibility of combining a high-field closed-bore MRI system, which would provide superior image quality, with this efficient means of switching between scanning and performing surgery.

Adaptations of Traditional Magnetic Resonance Imaging System Magnets

In addition to the custom MRI system magnets specially designed for intraoperative use, some iMRI systems incorporated preexisting low-field or high-field scanners into a modified OR environment. Horizontal gap systems, such as the Hitachi AIRIS I and II (Hitachi Medical Systems, Twinsburg, OH, and Kashiba, Chiba, Japan) and the Magnetom Open Viva (Siemens Medical Solutions, Erlangen, Germany), consisted of traditional low-field scanners with a single, integrated MRI-compatible OR table (Fig. 120-3). The table was modified so that it could rotate the patient’s head away from the magnetic isocenter and beyond the 5-Gauss (5G) line, where standard surgical instruments could be used. These iMRI systems were an extension of work initially done at the University of California, Los Angeles (UCLA), which first suggested that in addition to brain biopsies¹⁰ and transsphenoidal surgeries¹⁴ performed beyond the magnet’s 5G line, full craniotomies and surgeries could be safely conducted there as well.¹⁵ Monitoring and surgical devices that were needed within the isocenter zone (e.g., anesthesia equipment, electrocardiogram leads, and tools for interventional procedures) needed to be magnetic resonance compatible, whereas routine surgical instruments could be used in the outer zone. The patient transfer between imaging and surgery took only a few minutes, and sometimes the magnet was shared as a diagnostic MRI unit as well, thereby reducing overall costs. Despite the low-field nature of these systems (0.3 T), many investigators who have used them have reported their definite utility in the resection of gliomas and pituitary tumors and have even argued that they may be comparable to their high-field counterparts.^16,17

FIGURE 120-3 Siemens Magnetom Open Viva intraoperative magnetic resonance imaging (iMRI) system. This is a vertical-field 0.2-T open MRI system. Surgery is performed in the fringe field, where the weak magnetic field (<5 Gauss) does not interfere with standard operating room instruments and equipment.

(From Bergsneider M, Sehati N, Villablanca P, et al. Extent of glioma resection using low-field [0.2 T] versus high-field [1.5 T] intraoperative MRI and image-guided frameless neuronavigation. Clin Neurosurg. 2005;52:389-399.)

High-field systems such as the BrainSUITE (BrainLAB, Feldkirchen, Germany) use the same rotating MRI system OR table but with conventional 60- or 70-cm closed-bore magnets, Magnetom Sonata Maestro Class and Magnetom Espree (Siemens Medical Systems), respectively (Fig. 120-4). These 1.5-T systems afford the same degree of patient access and ease of transition between the magnetic and surgical fields while providing superior imaging quality and advanced acquisition techniques, such as fMRI, diffusion-weighted imaging (DWI), DTI, magnetic resonance angiography (MRA), magnetic resonance venography (MRV), and magnetic resonance spectroscopy (MRS).¹²

FIGURE 120-4 BrainSUITE intraoperative magnetic resonance imaging (iMRI) system. This iMRI system uses a traditional high-field 1.5-T scanner in conjunction with a modified, rotating operating table. The table facilitates the movement of the patient’s head from the isocenter of the magnet to a point outside the 5-Gauss (5G) line, allowing the transition between operating and imaging to occur within a few minutes. Once outside the 5G safety line, all traditional neurosurgical procedures can be conducted with standard MRI-incompatible instruments.

(Image courtesy of BrainLAB AG, Feldkirchen, Germany.)

Other innovations in modifying the OR suite to accommodate intraoperative imaging included the rail-mounted system (Fig. 120-5), which was first implemented in Calgary, Canada (IMRIS, Calgary, AB, Canada). This system emphasized the preservation of the traditional neurosurgical operating theater, arguing that many compromises were made to accommodate preexisting iMRI systems, including the need for custom magnetic resonance–compatible surgical equipment, restricted patient access, and the need to transport the patient within the OR or to an adjacent imaging room. Storing the magnet in a separate room not only permits use of the OR as a traditional neurosurgical OR but also allows the MRI unit to be used in other surgical suites or for diagnostic purposes. The 1.5-T closed-bore magnet can be moved from its resting site outside the OR into position and ready to image in less than 90 seconds. All magnetic resonance–incompatible surgical tools and equipment are moved outside the 5G line and powered down during imaging. This transition has been completed in as short a time as 10 to 12 minutes.³

FIGURE 120-5 Rail-mounted intraoperative magnetic resonance imaging (iMRI) system. The ceiling-mounted magnet can be moved in or out of the operating room in less than 90 seconds. By transporting the magnet into the operating room only when imaging is conducted, this iMRI system allows for the use of the traditional neurosurgical environment and surgical instruments without sacrificing imaging quality. In addition, the ceiling rails allow the deployment of the iMRI system in different operating theaters, helping to diffuse overall costs through interdepartmental utilization.

(Image courtesy of IMRIS, Calgary, AB, Canada.)

Aside from the original concept of conducting intraoperative imaging in the same time and space as conventional MRI, there have been many twin operating theater models in which the MRI system is located in an adjacent but otherwise separate room. This setup was first described by Tronnier and colleagues¹⁰ and was actually developed soon after the original double-doughnut model. The clear advantages of having a specialized MRI suite apart from the operating theater include being able to use MRI-incompatible surgical tools and equipment, having full patient access in a traditional surgical environment, and being able to use the MRI equipment in multiple ORs and departments, thereby greatly reducing overall costs.^3,16 In one such 3-T system, the patient is manually slid onto a modified MRI docking table, using a custom tabletop transfer system, and is subsequently moved down a sterile corridor into the MRI suite. The table docks with the MRI scanner, and imaging can then proceed. The portable anesthesia system that maintains sedation while the patient is away from the OR and the physiologic monitoring equipment are both MRI compatible.³

Glioma Resection

Much of the work with iMRI has focused on its ability to facilitate glioma resection. As was previously discussed, traditional stereotactic neuronavigation relies on fixed preoperative MRI scans of the lesion and its surrounding anatomy. The reliability of these images slowly diminishes throughout the surgery as CSF loss, surgical maneuvers, patient positioning, edema, and tumor resection all progressively alter the environment of the brain, producing increasing image inaccuracy with time (Fig. 120-6).^4–811 Neurosurgeons have been cognizant of this phenomenon of “brain shift” during volumetric stereotaxis since it was first reported by Kelly and associates in 1986.¹⁹ Although shifts of the surface cortex could be compensated for through the visualization of superficial structures and were appreciable by previously available intraoperative navigation, subcortical shift was poorly understood.⁸

FIGURE 120-6 Screen capture image taken from the BrainLAB VectorVision neuronavigation LCD monitor after intraoperative magnetic resonance imaging (iMRI) updating and reregistration, demonstrating the virtual pointer that identifies the contrast-enhancing tumor margin. Note the degree of brain shift (right upper image) that would not have been apparent had the preoperative image data set still been used.

More recently, however, with the advent of iMRI and the resultant ability to better assess shifts of cortical and subcortical matter in patients within the OR, investigators have begun to quantify the magnitude and direction of brain deformation under surgical conditions. Shifts of up to 5 cm have been reported on the cortical surface and of up to 3 cm within the deep brain tissue during glioma resection, with great interpatient variability.^5–8 The inaccuracy of stereotaxis is made apparent during glioma removal when the surgeon appreciates the discordance between the neuronavigation information and the appearance of the actual resection cavity. The surgeon’s inability to ascertain precisely the extent to which subcortical structures are shifted results in the overresection or, more often, underresection of lesions. With the use of iMRI, however, the neuronavigation system can be updated to agree with the intraoperative anatomy with an accuracy to within a couple of millimeters, even when using low-field systems.^8,11 Reregistering the patient’s data with the navigation system using high-field iMRI has been shown to improve the consistency of intended gross-total resections (GTRs) significantly, especially with gliomas larger than 40 cm³ (Fig. 120-7).¹¹ Although other surgical adjuncts, such as intraoperative ultrasonography, intraoperative CT, and mathematical models, have all been used to assess and even predict brain deformation, these methods still lack the resolution and compensatory ability to illustrate the changing brain structures adequately during neurosurgery.^4,7

FIGURE 120-7 A, Graphic comparison of the mean percentages of tumor resection in each of four randomized surgical groups: (1) conventional operating room (OR) with standard neuronavigation; (2) 0.2-T low-field iMRI; (3) high-field 1.5-T iMRI; and (4) high-field 1.5-T iMRI with intraoperatively updated neuronavigation. The 1.5-T iMRI with updated neuronavigational reregistration produced the highest mean and lowest variability in percentage of tumor removal. B, Comparison of preoperative (red circles) and postoperative (purple circles) tumor volumes (in cm³) in each group. Again, the 1.5-T iMRI with reregistration allowed the best and most consistent results. Error bars indicate 1 standard deviation of the mean.

(From Bergsneider M, Sehati N, Villablanca P, et al. Extent of glioma resection using low-field [0.2 T] versus high-field [1.5 T] intraoperative MRI and image-guided frameless neuronavigation. Clin Neurosurg. 2005;52:389-399.)

The benefit of rapidly updatable, near real-time anatomic imaging went hand in hand with the ability of iMRI to detect residual tumor, its boundaries, and its proximity to eloquent brain cortex in the surgical treatment of glioma. During the first decade of implementation, iMRI has been used most frequently after the neurosurgeon has determined that maximal tumor resection has been achieved. Although the results have varied greatly among groups, probably related to patient selection bias and different surgical philosophies at various institutions, evidence of residual resectable glioma on intraoperative scans at this point has been reported in 33% to 92% of cases.^{1,11,16,20–25} There are several possibilities for why such a large percentage of surgeries end with residual tumor present. For instance, the visual and textural similarity between common brain tumors, such as low-grade gliomas, and normal brain makes it difficult to distinguish pathologic tissue from healthy cortex during resection.²² In addition, the more malignant tumors are often characterized by microinvasion of the surrounding brain parenchyma, including eloquent areas, making complete resection without subsequent neurological sequelae very difficult, if not impossible. The inability of stereotactic navigation to reliably delineate tumor-brain margins because of dynamic brain shift throughout surgery can cause neurosurgeons to perform less aggressive resections to avoid postoperative deficits, especially if preoperative imaging indicates tumor proximity to brain regions controlling eloquent function. However, iMRI allows the neurosurgeon to better evaluate the progress of the resection, identify critical structures, and maximize more confidently the safe removal of tumor without inadvertently compromising neurological function.¹²

In the published literature, evidence of glioma remnants visualized on intraoperative images influenced surgical decision making most of the time, with additional resection being advocated in 50% to 92% of cases based on iMRI findings.²³ Nimsky and associates⁴ published a synopsis of patients treated during a 5-year period within their low-field 0.5-T iMRI suite. An analysis of the 95 patients with gliomas in this study revealed that the authors achieved initial GTR in 37% of their patients. In 38% of the remaining patients who had undergone incomplete tumor removal, resection was expanded based on iMRI findings, increasing the rate of total resection from 87% to 100% of patients with World Health Organization (WHO) grade I astrocytomas (n = 15) and from 25% to 56% in those who had grade II astrocytomas (n = 32). There were very few patients with high-grade gliomas whose resections were extended based on iMRI findings (n = 5), and this produced only a minimal benefit, increasing the rate of GTR from 42% to 47% in patients with grade III tumors and from 21% to 24% in those with grade IV tumors. Overall, complete tumor resection was improved from 37% to 51% in this study. The large benefit realized with iMRI in low-grade gliomas was further corroborated by Schneider and colleagues²⁶ who found that the use of a 0.5-T iMRI system increased the percentage of resected low-grade gliomas having less than 10% residual tumor volume from 33% to 92%. These findings suggest a significant benefit for iMRI in low-grade glioma surgeries, with only marginal benefit in resection of high-grade gliomas.

Other studies, however, have shown a significant utility for intraoperative imaging in patients with malignant (WHO grade III and grade IV) neoplasms, reporting that the use of low-field iMRI expanded resections of high-grade gliomas in 60% to 63% of patients.^16,24 Some groups have reported that the rate of GTR improved from between 6.5% and 36.6% to between 35.5% and 75.6% of cases when iMRI was used.^25,27 The inconsistency of data supporting use of iMRI in the case of high-grade gliomas may be the result of the varying degrees to which these more microinvasive tumors infiltrate surrounding eloquent brain tissue. The involvement of tumor in brain areas controlling eloquent function, which would result in neurological deficits if violated, frequently prohibits the aggressive resection of these neoplasms. Yet, the rate and degree in which eloquent cortex is involved may vary from one patient group to another, and this information is not regularly reported. Further randomized prospective studies may help to indicate the tumor conditions in which iMRI will be beneficial in the removal of high-grade gliomas.

Since the initial success in improving glioma resection through the use of low-field iMRI, the enhanced imaging quality and advanced acquisition techniques of high-field iMRI have marked these more sophisticated systems as the future generation of intraoperative imaging. In addition to the obvious benefits of faster scanning and superior image quality, the high-field magnet provides the ability to invoke other intraoperative imaging modalities such as fMRI, DTI, and MRS. Nimsky and colleagues²⁸ reported on their experience of using DTI intraoperatively with a high-field 1.5-T iMRI system to track pyramidal white matter fibers during resection of gliomas near motor areas. After obtaining routine preoperative images, including T1- and T2-weighted sequences for standard anatomic visualization, they included DTI and fMRI for pyramidal tractography and isolation of motor regions. The functional information was subsequently integrated into a three-dimensional magnetization-prepared rapid acquisition gradient-echo (3D MP RAGE) data set for intraoperative neuronavigation.¹⁸ The precentral gyrus, as identified by fMRI, was used as one of the seeds for fiber tracking in their knowledge-based dual region-of-interest (ROI) approach. The second arbitrary seed they used to determine the path of the pyramidal tract was located in the posterior parts of the internal capsule and the genu. From these two ROIs, tracking took place in both retrograde and orthogonal directions relative to the principal diffusion vector until the fiber tract was delineated. This process took only about 1 to 2 minutes and could thus be performed intraoperatively without too much concern for increased operating time. Just by employing the updated DTI and anatomic images alone, resection was extended in 6 of their 19 patients (32%). Postoperative analysis of preoperative and intraoperative fiber tracking further supported the use of iMRI-DTI in this scenario because they experienced pyramidal fiber shifts of up to 8 mm, with only 3 of 19 patients (15.8%) experiencing no detectable shift. In addition, no new neurological deficits were seen postoperatively.²⁸ Although further multicenter studies are necessary to evaluate the efficacy of this model on a larger scale, having such a system appears to be beneficial in improving resection of gliomas near eloquent cortex and fiber tracts.

Given the costs, increased operating time, and technique modifications needed to accommodate this new technology, it is appropriate to ask whether such measures to maximize glioma removal will actually improve patient outcome. To address the inference that increasing the degree of resection using iMRI may lead to improved patient survival, some groups have directly compared the outcomes of patients undergoing glioma removal with and without the aid of iMRI. Claus and associates²⁹ described their 6-year experience with 156 patients with low-grade gliomas who were operated on using iMRI and found that the age- and histology-adjusted mortality rates of these patients at 1, 2, and 5 years were significantly lower than those reported in the national databases. In addition, this group found that patients who underwent only subtotal resection instead of GTR were at a significantly increased risk for disease recurrence and death.²⁹ Other studies did not adjust for age or histology, but they also reported both significant and nonsignificant increases in survival time in patients who underwent GTR aided by iMRI compared with those who did not.^30,31

There is increasing evidence that imaging-documented complete resection of low-grade gliomas may improve patient survival time, despite the relative paucity of similar evidence in the more difficult to resect high-grade tumors. This has caused some to suggest that the potential for cure in patients with some WHO grade I and II gliomas through GTR may warrant the use of iMRI, whereas the need for iMRI in patients having more malignant gliomas may not be as clear.²² Regardless, because the current operative goal in almost all glioma patients is to resect as much of the tumor as possible without causing new postoperative neurological deficits, iMRI may serve as a critical tool, especially when the tumor borders or invades eloquent cortex, irrespective of its definitive effect on patient mortality. As such, with increasing availability, iMRI may someday become part of the standard of care for glioma surgery.³²

Brain Biopsy

During the past few decades, brain biopsy techniques have progressed from freehand sampling using CT to stabilized, MRI-based stereotaxis procedures.³³ A large literature review revealed that stereotactically guided brain biopsies yield diagnostic tissue about 91% of the time.³⁴ Nondiagnostic biopsies are often the result of inaccurate tissue targeting, small target size, small sample size, and attempting to biopsy areas with an increased T2-weighted magnetic resonance signal.³⁵ Many of these problems can be attributed to the use of static neuronavigation systems reliant on preoperative images that are unable to account for intraoperative brain shifts, as was discussed earlier. Although one can expect to observe a minimal amount of shift during brain biopsy compared with more extensive intracranial procedures such as large tumor resections, even small discrepancies in stereotactic and actual space can become problematic when targeting smaller lesions. Furthermore, after radiation treatment or infectious processes, it is difficult to distinguish the target of interest (e.g., tumor, active infection) from areas of postirradiation necrosis or gliosis in heterogeneous lesions showing contrast enhancement on T1-weighted magnetic resonance images, thereby complicating diagnostic sampling.³⁶ iMRI, with its ability to provide near real-time updates of the biopsy procedure, as well as aid in the functional discrimination of pathologic from normal brain tissue (through MRS or perfusion MRI), can address many of these problems.

In June of 1995, at Brigham and Women’s Hospital in Boston, a stereotactic brain biopsy was the first procedure to be completed in the iMRI environment.⁹ After positioning the patient within the Signa SP scanner, preoperative images were taken for biopsy planning, and a bur hole was subsequently made. Simply by using a Brookwalter arm, the biopsy cannula was fixed in place as the needle was passed through into the brain under direct real-time image guidance.³⁷ The double-doughnut open-MRI system allowed for repeated imaging of the needle’s trajectory as it continued through to the biopsy site, without needing to move the patient. By maintaining visualization of the needle throughout the procedure, entry into the target lesion and avoidance of critical structures could be confirmed.

Since then, other groups have modified their OR environments to conduct intraoperative imaging for brain biopsies without the use of equipment specifically designed for iMRI.^10,38 In these setups, trajectory planning took place within a sterile MRI suite. However, after imaging was performed, the patient was moved away from the magnetic isocenter, beyond the 5G line. In this magnetic “fringe field,” conventional surgical equipment, including drills and electrocautery, could be used. Nevertheless, because biopsy needle insertion took place within the magnet, MRI-compatible instruments necessary for guided tissue sampling were still required.^10,38 Although initial studies of iMRI-guided brain biopsies revealed diagnostic tissue in 97% to 100% of cases, the relatively small numbers of patients included made it difficult to compare these results directly with outcome data in series of patients undergoing more conventional frame-based and frameless stereotactic biopsies conducted without the aid of iMRI.^37–39

Prominent developments after the first iMRI-guided biopsies have continued to improve the reliability and diagnostic ability of this technology. Needle-stabilizing devices were soon developed because freehand biopsies have often been criticized for being inaccurate and potentially increasing the risk for acute intracerebral hemorrhage.³⁵ Currently, skull-mounted trajectory guides with three-point alignment algorithms help account for any patient head movements, maintain the desired trajectory, and stabilize the needle throughout the procedure.^33,40

Another advancement was the addition of MRS in high-field iMRI systems to facilitate the localization of pathologic areas and subsequent biopsy of diagnostic tissue. Stereotactic biopsy has often been used to distinguish recurrent neoplasm from postirradiation necrosis, given the similarity of these two findings on imaging scans.³⁸ In patients who have received external beam radiation for central nervous system tumor treatment, this becomes problematic because imaging studies cannot conclusively distinguish the two, and biopsy may not accurately sample the pathologic tissue within a large lesion that shows heterogeneous contrast enhancement. Through intraoperative MRS techniques, including turbospectroscopic imaging (TSI) and single-voxel spectroscopy (SVS), spectroscopic maps based on levels of cellular metabolites (e.g., choline and N-acetyl aspartate) can be superimposed on images of cortical anatomy, guiding the biopsy needle to pathologic tissue.^35,36,41 Diagnostic tissue was obtained in 100% of cases in two preliminary studies using intraoperative MRI or MRS.^35,36

Initial studies of iMRI-guided brain biopsies have shown marked reductions in morbidity and superior rates of diagnostic tissue sampling compared with conventional stereotactic navigation systems.^39,42 In addition to being able to plan the biopsy path in nearly real time to avoid critical anatomic structures, the ability of iMRI to detect intraoperative complications contributes strongly to reducing patient morbidity. This is especially true for intracranial hemorrhage, which has been reported to occur in 0% to 11.5% of stereotactic brain biopsies.³⁷ Initially, bleeds were assessed by observing changes in serial images that were consistent with blood collection, although more recently, with high-field systems, a combination of T2-weighted half-Fourier acquisition single-shot turbo spin echo (T2-HASTE), gradient echo (GE), and fluid-attenuated inversion recovery (FLAIR) techniques has been shown to sensitively detect hyperacute hemorrhage.⁴³ Through immediate detection and evacuation of bleeds during surgery, rather than later in response to clinical deterioration, neurological sequelae can be avoided.

Transsphenoidal Resection of Pituitary Adenomas

Although the first transsphenoidal approach for pituitary adenoma removal was performed more than a century ago,⁴⁴ it was relatively recently, with the development of adjuvant technologies, that it became the method of choice in the resection of pituitary tumors.⁴⁵ Innovations such as intraoperative fluoroscopy, the operating microscope, frameless stereotaxy, and the neuroendoscope have all aided in improving navigation to the tumor bed and visualization of the tumor to maximize resection.^45–47 However, these advances were not without their limitations. Although fluoroscopy allows for rapid real-time imaging, its visualization is limited to the bony structures of the sella, and it is unable to aid in tumor detection.⁴⁸ The operating microscope greatly increases visibility within the surgical field, but it is limited by angles of viewing, and microscopic visualization alone does not permit appreciation of the entire sella and its surrounding regions.⁴⁵ The phenomenon of brain shift interferes with the accuracy of frameless stereotactic systems of neuronavigation (as described earlier in the case of glioma resection), and these systems cannot be updated intraoperatively to identify remaining tumor, thus decreasing their utility. Finally, despite the drastic improvement that endoscopy has provided relative to the surgical microscope alone, it requires direct visualization to confirm the presence or absence of tumor and does not allow one to see beyond the surface anatomy.⁴⁵ This becomes especially troublesome in the case of tumors with suprasellar and parasellar extension because focused removal requires appreciation of structures located beneath the surgical field of view. Given these limitations, it is not surprising that complete removal of pituitary adenomas remains a problem, with studies suggesting that this is achieved in only about 65% of macroadenoma cases.⁴⁹

iMRI brings a distinct advantage to the previous armament of tools for transsphenoidal navigation and neuroimaging. Because it is unhindered by a need for direct visualization of the obscured, ever-changing resection cavity, iMRI can provide images of tissue during an operation that enable assessment of residual tumor. Serial updates allow the neurosurgeon to localize remnant tumor while accounting for brain shift throughout the surgery and to direct resection while preserving critical native structures (e.g., the optic chiasm and cavernous sinuses). Furthermore, dynamic image sequences help distinguish normal pituitary tissue (fast contrast enhancement) from adenoma (slow contrast enhancement) and hemorrhage (heme-sensitive gradient echo).^46,48

Transsphenoidal pituitary macroadenoma surgery performed with low-field iMRI systems revealed residual tumor in 20% to 66% of patients after initial resection, with some investigators reporting the ability to achieve complete removal in 75% to 100% of cases with the use of iMRI.^4,45–47 In a study using high-field iMRI, Nimsky and colleagues⁵⁰ found that with intraoperative scanning, they were able to increase their rate of complete pituitary tumor removal from 58% to 82%, despite selecting for patients with macroadenomas and large suprasellar extension. They argued that high-field iMRI is notably better than its low-field equivalent for pituitary adenoma surgery because it can detect tumor remnants as small as 3 to 4 mm in maximal diameter and is not subject to any of the artifact or image quality problems found in the low-field systems. The importance of magnet strength in detecting small tumor pieces is supported by other studies that argue the superiority of 3-T MRI relative to standard 1.0- to 1.5-T magnets in detailing the invasion of pituitary adenomas into adjacent structures, during preoperative planning.⁵¹ In addition to the immediate benefits of guiding and improving resection, iMRI provides instantaneous feedback regarding the success of the operation. This allows the surgeon to determine on the day of surgery whether further management is required, rather than waiting the standard 2 to 3 months for artifact-free postoperative scans to decide among surveillance, radiation therapy, or secondary transcranial surgery.^4,50

Several other groups reported superior rates of complete tumor removal while using iMRI to guide transsphenoidal resection.^4,45–48,50 However, the inclusion criteria among these studies varied greatly, with some choosing nonfunctioning macroadenomas with suprasellar extension, whereas others did not specify the hormonal status or amount of tumor extension into neighboring brain regions. This makes objective comparison of surgical outcomes difficult. Furthermore, a study conducted by Fahlbusch and associates⁵² assessing the use of iMRI in the transsphenoidal resection of growth hormone (GH)-producing pituitary macroadenomas found that even high-field iMRI could not detect tumor remnants in all cases, as confirmed by persistent endocrine abnormalities in patients, despite negative findings on intraoperative scans. Consequently, it may be too early to conclude whether the reported imaging outcomes translate into truly beneficial clinical outcomes that can be compared with national statistics derived from studies with more than 10 years of follow-up.⁴⁹ Given the relatively few studies using iMRI specifically for pituitary tumors and their relatively short follow-up periods, the degree to which iMRI guidance will improve the rates of surgical cure in pituitary adenomas has yet to be determined.

Intractable Epilepsy

Although surgery has been a treatment option for pharmacoresistant epilepsy for more than 50 years, it was only recently that a randomized controlled trial⁵³ investigated the efficacy of surgery compared with medical management. The study found that surgical intervention for temporal lobe epilepsy allowed 58% of patients to be seizure free after 1 year, compared with only 8% of the medically managed group. This outcome, in conjunction with other reports of increased seizure-free intervals after surgery, has continued to support this invasive treatment option over the years. Surgical techniques available for intractable epilepsy include focal cortical resection, anatomic lobectomy, lesionectomy, corticectomy, and hemispherectomy and its variants.⁵⁴ The actual procedure performed depends on the location of the seizure focus and characterization of the epilepsy syndrome involved. For example, patients with epilepsy secondary to a specific seizure focus, including neoplasms and vascular malformations, undergo focal cortical resection to remove the epileptogenic lesion; alternatively, mesiotemporal lobe epilepsy patients with hippocampal sclerosis benefit from amygdalohippocampectomy or en bloc temporal lobe resection.⁵⁴ A review of previous literature reveals mounting evidence that the moderate to high recurrence rate (20% to 60%) after surgery in seizure syndromes such as temporal lobe epilepsy may be a result of the inadequate resection of implicated structures, such as the amygdala and hippocampus, with freedom from seizures being correlated with more complete and accurate resection.^55,56 Because iMRI had proved useful in surgeries in which the extent of resection was critical to patient outcome, it appeared reasonable to extend its application to epilepsy surgery.

The few studies that have investigated the use of iMRI in epilepsy surgery have looked at the treatment of distinct seizure pathologies. Overall, they report the utility of intraoperative imaging at helping navigate through cortex while accommodating for brain shift as well as aiding in the detection and resection of residual epileptogenic structures or lesions.^55–58 Kaibara and coworkers⁵⁶ reported that although iMRI was particularly useful in the more selective amygdalohippocampectomy performed in a small subset of their patients, intraoperative imaging allowed detection of residual hippocampus or amygdala in 50% of all patients, suggesting that even with the more aggressive anterior temporal lobectomy, critical structures implicated in seizure production may be left behind. Among their patients, 93% who underwent surgery with iMRI were seizure free at a mean follow-up interval of 17 months, supporting their surgical outcomes. Schwartz and associates⁵⁵ published similar results in a small study in which 100% of their patients were seizure free at 10 months. Other studies reported the use of iMRI in resecting neoplasm-related seizure foci but described more moderate benefits of this technology in seizure control.^57,58

In addition to the obvious benefit of iMRI in removing seizure-producing lesions, it was also shown to guide the elimination of nonlesional epileptogenic foci, undetectable by traditional imaging modalities. Through iMRI-guided placement of depth electrodes, neurosurgeons are able to customize the resection by mapping out the precise locations of aberrant neuronal firing.⁵⁷ It also allows for more confident extension of the resection in both lesional and nonlesional cases, if electrocorticography reveals residual ictal activity. This technique has been reported previously with other imaging modalities such as CT and preoperative MRI-based stereotaxis⁵⁹; however, these earlier systems were incapable of performing real-time anatomic updates to account for brain shift in the operative setting. Thus, iMRI may serve as a quality control enhancement in the removal of both lesion-based and nonlesional seizure foci.

Other Applications

Given the broad range of procedures that require precise localization and near real-time knowledge of intracranial anatomy during neurosurgical procedures, it is not surprising that iMRI has been used for many different indications. Less frequent applications of iMRI include use during intracranial cyst drainage, abscess evacuation, deep brain stimulator placement, and assessment of vascular compromise using MRA in high-field systems.^1,4,9,41,60