We used the twin neurosurgical-MRI suite for 21 patients. Twenty-six iMRI examinations were carried out, of which 3 were immediately presurgical (only children), 9 were intrasurgical (partial head closure before iMRI), and 14 were immediately postsurgical (head fully closed, but patient still anesthetized and sterilely draped). Five patients had two iMRI procedures during the same intervention, and three had additional tumor resection immediately after iMRI, of whom two had partial head closure and one was already completely closed (Table 2).
All patients were anesthetized, monitored with MRI-compatible electrodes installed dorsally (n = 19) or ventrally (n = 2), and positioned without restraints for surgery, with the head held in a three-pin MRI-compatible headholder. To avoid any electrical loop that could lead to skin burns in a contact zone during MRI, each limb was draped and electrically isolated from the rest of the body. Once the neurosurgical procedure was thought to be completed, patients were temporarily (n = 9) or completely (n = 14) closed, depending on the surgeon's assessment. For partial closure, a large gelatin foam sponge (Gelfoam; Pharmacia & Upjohn Co., Kalamazoo, MI) covered the exposed brain, two or three stitches were placed to stretch the dura mater, and two or three large stitches were placed on the scalp, the whole partial closure being done in less than 5 minutes. The reopening of such a partial closure would take less than 2 minutes.
The surgical wound was protected by sterile compresses and adhesive tape. The drapes were cut around the patient's head and body. The BrainLAB reference star-holder was detached from the headholder, and the whole headholder system and head were repositioned above the plane of the surgical table to allow insertion of the mobile table extension. The nonsterile MRI flexible surface coils were then fixed on the patient's head with regular tape.
Once the surgical team was ready, the motorized surgical table on rails was moved to the MRI facility. A trained senior neuroradiologist interpreted intraoperative images immediately after acquisition. The surgeon evaluated whether additional resection was appropriate when residual tumor was clearly delineated. If not, when necessary, patients were redraped, and a complete head closure was performed (n = 9). If additional surgery was indicated, the operating field was reopened, and tumor resection was completed. If needed, the additional tumor resection was followed by a second iMRI examination (n = 2).
All involved personnel were thoroughly trained to avoid accidents owing to ferromagnetic objects entering the MRI facility; no such events have occurred. Similarly, no accidents have occurred during patient transfer between the OR and MRI room. We carefully avoided placing wire loops on patients' skin to prevent burns caused by electromagnetically induced eddy currents (also named Foucault currents). Nevertheless, one patient experienced a second-degree burn of the intergluteal sulcus that we attributed to conductive moisture caused by local sweating. The patient had been placed on an absorbent pad with a plastic underlay that we suspected had caused excessive sweating. This kind of pad was not used subsequently. Similarly, another patient placed in the ventral decubitus position experienced chest burns attributed to the covering on the cushions that were used. We ceased using these types of cushions, and no additional burns were observed. Indeed, conducting perspiration induced by prolonged contact between skin and polyacrylate (or vinyl) materials (synthetic tissues and drapes) have been responsible for superficial burns caused by eddy currents. Discontinuing their use has resulted in the disappearance of such a drawback in subsequent patients so far.
Every planned iMRI procedure was successfully completed. The technical dysfunctions that did occur were minor and were attributed to the learning curve required to master the technical tricks and widgets of the OR-MRI suite. A senior MRI technician was present in the suite for only the two initial procedures (Table 3).
Each type of technical problem that we encountered occurred only once or twice, except for the head being too high in the MRI tunnel (four times). Any metal artifact that we encountered did not disturb the MRI interpretation, and the two incidents that did occur did not lead to any loss of time because the patient's head did not need to be repositioned. Other MRI-related dysfunctions were solved by online collaboration with the radiological team. In the beginning, the surgical table frequently became blocked, but this did not cause significant delay. A technical check of the table revealed an internal problem that was easily solved, and no further table blockage occurred.
For the last 10 patients, we recorded the time taken for each phase of the iMRI procedure (Fig. 7). The time required to perform the whole procedure ranged from 58 minutes in Patient 15 to 129 minutes in Patient 14, in whom three dysfunctions occurred, accounting for 58 minutes of delay (Table 3). On average, the procedure took 78 ± 20 minutes, with an average of 34.1 minutes for the MRI itself. Avoidable delays occurred when the OR had to wait for the MRI room to be converted to surgical conditions (Table 3); this happened in four cases, three of them in the second half of this series. Delays were attributed to surgery taking less time than projected, calling for MRI room preparation less than an hour before the end of the surgery, or the radiologist's program delay with the 3-T MRI in research mode. These issues account for the wide time range (10–43 min) observed for the transfer preparation.
The impact of iMRI on the surgical assessment for this series of patients was also prospectively studied (Table 4). One patient who benefited from epilepsy surgery was excluded from this aspect of the study because only multiple subpial transections based on electrocorticography were performed. Perioperative iMRI interpretation concluded that tumor or dysplastic cortex removal or lobar disconnection was complete in 14 of 20 cases, including all intraventricular tumors (n = 2) and lobar disconnections (n = 2), 2 of 3 meningiomas, 2 of 3 pituitary adenomas, 5 of 6 gliomas, and 1 of 3 metastases. The iMRI scans of one meningioma, one vestibular schwannoma, and one pituitary macroadenoma were interpreted intraoperatively as near total, partial, and subtotal, respectively, and no additional resection was performed. For these three cases, the surgeon considered that additional resection could be not performed with a favorable benefit-to-risk ratio.
In 3 of 20 cases, the iMRI interpretation led to a modification of the surgical assessment with additional tumor resection. One patient (Patient 14) (Fig. 8; Table 2) who presented with a Grade III glioma in the precentral region had an iMRI scan interpreted as near total resection. The removal was completed with a resection and was shown to be total by a 24-hour postoperative MRI. The residue histology showed slight tumor infiltration. The two patients with intra-axial metastasis had a near total tumor resection followed by a post-iMRI residue resection with a second iMRI perioperatively interpreted as total (Patient 5, Fig. 9; Patient 15, Fig. 10). In both cases, the surgeon macroscopically identified the tumor residue.
The surgical goals and surgical assessments were collected before the iMRI scan was done and were compared with iMRI interpretation (Table 5). When the surgeon wanted to perform a total tumor removal or lobar disconnection (n = 18), he considered the resection or disconnection total in 17 cases, 4 (24%) of which were actually interpreted as near total on iMRI. One of these patients (Patient 6) (Table 2) had no further surgery. The three other patients were those described in the preceding paragraph.
Spatial resolution is based on the theoretical linearity of the relationship between the strength of the magnetic field and the signal-to-noise ratio (SNR). Without changing any other conditions, the SNR increases as the field strength increases. This allows a choice between shorter acquisition time for the same IQ and improved IQ, in terms of spatial resolution, without increasing acquisition time (3,4,21).
For surgical purposes, spatial resolution is the most important parameter affecting the quality of surgical resection. A better SNR also allows a better recognition of deep gray, thalamic, and subthalamic nuclei (15). Even if the gain obtained by shifting from 1.5- to 3-T is not exactly double in terms of time and SNR, the IQ of the brain, the nerves, and the vessels is noticeably improved (11). Other advantages of high-field-strength imaging are the ability to acquire three-dimensional isotropic images (voxels of equal size) in a reasonable acquisition time and the ability to study a volume in any possible plane with the same spatial resolution.
The choice of a 3-T MRI magnet for coupling imaging and surgery was based on the assumption that over the next 7 years, day-to-day imaging practices in the university hospital setting will gradually move from 1.5- to 3-T (19). Our aim is to offer IQ inside the OR similar to that available outside it.
Using a 3-T MRI scanner is associated with some well-known disadvantages that can be managed by choosing the appropriate technical options and adapting specific procedures: 1) increasing radiofrequency energy deposition (specific absorption rate): the use of gradient echo, inversion recovery, and other new sequences with hyperechoes; 2) altering the contrast: namely, the increase of T1 with lesser contrast between white and gray matter in the T1 ponderation; 3) stronger susceptibility effect and metallic artifacts: the use of new coils and acquisition parameters with algorithms like BSENSE (sensitivity encoding), BmSENSE (modified SENSE), BPILS (parallel imaging with localized sensitivities); this stronger susceptibility can also be an advantage in detecting hemosiderin deposition and bleeding lesions containing deoxyhemoglobin; 4) more moving artifacts: the use of adapted sequences and preference for gradient echo sequences to spin echo sequences; 5) larger chemical shift, e.g., use of sequences with fat saturation.
The iMRI procedures that we performed caused two second-degree skin burns that healed without additional sequelae. The single case of wound infection was not attributed to the iMRI procedure because the infection was due to a secondary reopening of the wound caused by a transient movement disorder. Because no other complications associated with the iMRI occurred, we conclude that use of the twin neurosurgical-MRI suite at 3-T is safe, and intraoperative procedures should continue.
The design of the OR-MRI suite allows normal neurosurgical procedures to be performed without any MRI-linked restrictions. The high standards set for surgeon comfort are counterbalanced by the complexity of a multiphase transfer procedure. Indeed, eight different types of transfer-associated dysfunctions were encountered, which disturbed 12 of the procedures. Six types of dysfunctions occurred no more than twice: blocked transfer table (resolved by technical inspection of the table), blocked MRI table (unrelated to the iMRI procedure), unplugged coil (the disconnection between the plug and the coil wire was systematically checked), coil position artifacts (owing to overlap or close contacts between the coils), metal artifacts (the removal of the star-holder was double checked), and MRI software bug (unrelated to the iMRI procedure). The two remaining dysfunctions were a prolonged waiting period (14–30 min) between the end of the surgery and the availability of the MRI room and the head being positioned too high to fit into the MRI tunnel. The prolonged waiting period depended on the anticipated remaining surgical time and the constraints of the neuroradiological team. This potential waiting period remains the most unpredictable aspect of the procedure. We solved the head position problem by disconnecting the headholder from its base if the head was too high. Regardless of the type of dysfunction that occurred, every planned iMRI procedure was completed successfully. We consider these dysfunctions to be part of the learning process because, for the next 100 patients, the main problem that occurred occasionally was the prolonged waiting period.
Historically, iMRI has been developed in conjunction with neuronavigation systems designed to be used within the flow of surgery with MRI-compatible surgical rooms and tools. The goal was to obtain live brain images during surgery; therefore, it was crucial to have a simple and fast imaging procedure that allowed repeated iMRI. Currently, this need is filled by various neuronavigation systems and iMRI can be used as a quality control system that allows for additional tumor resection. In our experience, repeating an iMRI scan was rarely needed and was only done twice. The iMRI procedure is time-consuming, which is the main drawback of the suite that we developed; the procedure adds an average of 78 ± 20 minutes to the normal surgical time. The average time needed for MRI is 34 minutes, and the average total transfer time is 43 minutes, with a minimum of 32 minutes (Patients 15 and 20). However, considering the potential added benefit for the patient, this additional time seems reasonable. Interestingly, Schwartz et al. (14a) reported that a craniotomy performed under iMRI conditions lasted an average of 1 hour and 42 minutes longer than under conventional conditions, owing mainly to the setup in the double doughnut magnet. Similarly, Martin et al. (7) calculated that tumor resection performed with iMRI took approximately one-third longer than in a conventional operating room. Finally, the use of the PoleStar N10 was calculated to lengthen surgical time by 1.7 hours compared with conventional surgery (13).
General considerations, issued from routine diagnostic imaging, about the benefits of increasing the basic magnetic field B0, fully applied to the obtained intraoperative images. Mainly, a higher SNR allows either shorter acquisition time at constant IQ or improved IQ (mainly higher spatial resolution) in a similar acquisition time. We were committed to preserving a similar IQ of iMRI when compared with routine diagnostic imaging. Sensitivity encoding (SENSE) parallel imaging option was used, resulting in significant time savings. Overall, in this study, the intraoperative IQ was rated similar to that of conventional imaging except in the deepest areas, where a slight loss of SNR was observed because of the use of the surface coil instead of quadrature bird cage head coils.
In this preliminary study, we decided to perform iMRI in every brain surgery except brain biopsies. Indeed, the time needed for iMRI would have been as long as the time needed for the biopsy, which seemed unacceptable to us. This study included only three pituitary adenomas that required no further surgery after iMRI. We could not discern any trends with regard to the usefulness of 3-T iMRI for this particular pathology. Three patients (16.6%) undergoing tumor resection required an additional surgical removal. These results are in line with the lowest rate of required additional surgery reported in the literature, which ranges from 16% (13) to 67% (16). Analysis of the glioma surgeries shows that, of the six patients with glioma who underwent operations in this study, only one (16%) required further tumor resection, a rate lower than the 28% published by Nimsky et al. (8), but our series of gliomas is too small to reach statistical significance.
Surprisingly, two of the three patients with metastasis required additional surgery. Thus, metastasis accounts for the highest rate of additional surgery in our study, which is close to the highest rate of additional surgery reported in the literature (67%). This was unexpected because brain metastases are known to be well-delineated tumors. One possible explanation for this is that one of these patients had undergone gamma knife radiosurgery for the metastasis before surgery. This preliminary procedure could have made it difficult for the surgeon to identify the tumor margin. However, this group is also too small to allow us to draw any meaningful conclusions.
The interpretation of the 3-T iMRI is crucial to the neurosurgeon's decision-making process. Surgical information provided to the neuroradiologist facilitated interpretation of the iMRI. However, in five (21%) of 23 intra- and postsurgical MRI scans, the interpretation changed, illustrating how challenging interpreting iMRI can be. Nonspecific peritumoral enhancement after gadolinium injection is known to be difficult to interpret with lower-field iMRI (5). The accuracy in the definition of tumor remnant versus nonspecific enhancement is a major limit of iMRI regardless of the intensity of the magnetic field used; however, the 3-T iMRI should be able to overcome this limitation. At 3-T, intraoperative spectroscopic and perfusion studies of the margin of the tumor resection cavity are technically possible within a clinically acceptable period; we are currently evaluating the accuracy of these tools for defining the putative tumor remnant.
The OR-MRI suite at 3-T was developed by a multidisciplinary team. The concept of two independent surgical rooms, one with a 3-T MRI machine, separated by an air lock chamber that could be linked when iMRI is needed, proved to be safe and practicable, albeit time-consuming. The IQ allowed thorough analysis of the tumor resection margins, providing us with the tools needed to distinguish tumor tissue from nonspecific gadolinium enhancement. Our initial experience suggests that iMRI at 3-T is safe and feasible and can influence the extent of surgery for glioma and metastasis, but larger studies are needed to draw firm conclusions.
This work was supported by a grant from the Belgium Fond National de la Recherche Scientifique.
1. Black PM, Moriarty T, Alexander E 3rd, Stieg P, Woodard EJ, Gleason PL, Martin CH, Kikinis R, Schwartz RB, Jolesz FA: Development and implementation of intraoperative magnetic resonance imaging and its neurosurgical applications. Neurosurgery
2. Hadani M, Spiegelman R, Feldman Z, Berkenstadt H, Ram Z: Novel, compact, intraoperative magnetic resonance imaging-guided system for conventional neurosurgical operating rooms. Neurosurgery
3. Hall WA, Galicich W, Bergman T, Truwit CL: 3-Tesla intraoperative MR imaging for neurosurgery. J Neurooncol
4. Hall WA, Liu H, Martin AJ, Pozza CH, Maxwell RE, Truwit CL: Safety, efficacy, and functionality of high-field strength interventional magnetic resonance imaging for neurosurgery. Neurosurgery
5. Knauth M, Aras N, Wirtz CR, Dörfler A, Engelhorn T, Sartor K: Surgically induced intracranial contrast enhancement: Potential source of diagnostic error in intraoperative MR imaging. AJNR Am J Neuroradiol
6. Kollias SS, Bernays R, Marugg RA, Romanowski B, Yonekawa Y, Valavanis A: Target definition and trajectory optimization for interactive MR-guided biopsies of brain tumors in an open configuration MRI system. J Magn Reson Imaging
7. Martin AJ, Hall WA, Liu H, Pozza CH, Michel E, Casey SO, Maxwell RE, Truwit CL: Brain tumor resection: Intraoperative monitoring with high-field-strength MR imaging—Initial results. Radiology
8. Nimsky C, Ganslandt O, Von Keller B, Romstöck J, Fahlbusch R: Intraoperative high-field-strength MR imaging: Implementation and experience in 200 patients. Radiology
9. Pamir MN, Peker S, Ozek MM, Dinçer A: Intraoperative MR imaging: Preliminary results with 3 Tesla MR system. Acta Neurochir Suppl
10. Roessler K, Ungersboeck K, Czech T, Aichholzer M, Dietrich W, Goerzer H, Matula C, Koos WT: Contour-guided brain tumor surgery using a stereotactic navigating microscope. Stereotact Funct Neurosurg
11. Röttgen R, Haltaufderheide K, Schröder RJ, Lorenz M, Herzog H, Neumann F, Lehmkuhl L, Winter L, Felix R, Bruhn H: The effect of the field strength on standardized MRI of the brain to demonstrate cranial nerves and vessels: A comparison of 1.5 and 3.0 Tesla [in German]. Rofo
12. Schenck JF, Jolesz FA, Roemer PB, Cline HE, Lorensen WE, Kikinis R, Silverman SG, Hardy CJ, Barber WD, Laskaris ET: Superconducting open-configuration MR imaging system for image-guided therapy. Radiology
13. Schulder M, Liang D, Carmel PW: Cranial surgery navigation aided by a compact intraoperative magnetic resonance imager. J Neurosurg
14. Schulder M, Salas S, Brimacombe M, Fine P, Catrambone J, Maniker AH, Carmel PW: Cranial surgery with an expanded compact intraoperative magnetic resonance imager. Technical note. J Neurosurg
14a. Schwartz RB, Hsu L, Wong TZ, Kacher DF, Zamani AA, Black PM, Alexander E 3rd, Stieg PE, Moriarty TM, Martin CA, Kikinis R, Jolesz FA: Intraoperative MR imaging guidance for intracranial neurosurgery: Experience with the first 200 cases. Neuroradiology
15. Spiegelmann R, Nissim O, Daniels D, Ocherashvilli A, Mardor Y: Stereotactic targeting of the ventrointermediate nucleus of the thalamus by direct visualization with high-field MRI. Stereotact Funct Neurosurg
16. Staubert A, Pastyr O, Echner G, Oppelt A, Vetter T, Schlegel W, Bonsanto MM, Tronnier VM, Kunze S, Wirtz CR: An integrated head-holder/coil for intraoperative MRI in open neurosurgery. J Magn Reson Imaging
17. Steinmeier R, Fahlbusch R, Ganslandt O, Nimsky C, Buchfelder M, Kaus M, Heigl T, Lenz G, Kuth R, Huk W: Intraoperative magnetic resonance imaging with the magnetom open scanner: Concepts, neurosurgical indications, and procedures—A preliminary report. Neurosurgery
18. Sutherland GR, Kaibara T, Louw D, Hoult DI, Tomanek B, Saunders J: A mobile high-field magnetic resonance system for neurosurgery. J Neurosurg
19. Tanenbaum LN: Clinical 3T MR imaging: Mastering the challenges. Magn Reson Imaging Clin N Am
20. Tronnier VM, Wirtz CR, Knauth M, Lenz G, Pastyr O, Bonsanto MM, Albert FK, Kuth R, Staubert A, Schlegel W, Sartor K, Kunze S: Intraoperative diagnostic and interventional magnetic resonance imaging in neurosurgery. Neurosurgery
21. Truwit CL, Hall WA: Intraoperative magnetic resonance imaging-guided neurosurgery at 3-T. Neurosurgery
58[Suppl 2]:ONS-338–ONS-346, 2006.
22. Wagner W, Tschiltschke W, Niendorf WR, Schroeder HW, Gaab MR: Infrared-based neuronavigation and cortical motor stimulation in the management of central-region tumors. Stereotact Funct Neurosurg
23. Yrjänä SK, Katisko JP, Ojala RO, Tervonen O, Schiffbauer H, Koivukangas J: Versatile intraoperative MRI in neurosurgery and radiology. Acta Neurochir (Wien)
Jankovski et al. provided a comprehensive description of their early experience with a 3-T intraoperative magnetic resonance imaging (iMRI) unit. In the first part of their report, they provided a stepwise account of their conception, planning, and construction of an MRI suite designed so that the unit can be shared for diagnostic ambulatory patient scanning. They addressed important considerations including room layout, safety, ergonomics, surgical instrumentation, anesthesia modifications, and imaging protocol design. In doing so, they provided general insight into the multitude of factors that must be weighed for institutions considering new iMRI installations.
In the second part of their report, Jankovski et al. summarized their initial experience with iMRI for a variety of intracranial pathological lesions including meningiomas, colloid cysts, astrocytomas, pituitary tumors, and metastases. The number of patients is small and their experience is nascent: so far they reported that iMRI led to additional resection in three patients (14.3%). Interpretation is limited by the lack of uniform postoperative MRI assessment and long-term clinical outcomes. Perhaps more important than the numbers, they provided details about dysfunctions and delays that illustrate the learning curve and time required to incorporate this technology successfully into the operating room (OR). The most useful iMRI system is the one that gets the most use. A 3-T magnet may have some technical advantages over a 1.5-T magnet in terms of imaging. However, choosing a 3-T system introduces problems that may make the system less useful in practice as 3-T magnets are more expensive, have a larger footprint, and require more elaborate shielding and safety measures. The higher fixed costs of the machine and installation may increase the motivation for shared usage, which maximizes the opportunity to increase revenue from nonoperative scanning. In this study, Jankovski et al. found that shared usage came at the cost of reduced throughput for the surgeons. Intraoperative scans were delayed while the scanner was in use. It is notable that for many patients only a postoperative image was obtained, suggesting significant barriers (real or perceived) to obtaining “on-the-fly” scans. Practical use of a 3-T system is further encumbered by additional safety considerations. They noted two patients with thermal injuries owing presumably to current loops from skin-to-skin contact. The burn risk can be addressed by careful positioning and thorough inspection before each scan, although these procedures take additional time.
Overall, this study highlights the need for detailed cost-benefit analysis when one is considering 3-T versus 1.5-T technology for iMRI. Image quality is only one part of the equation. In the end, practical considerations were paramount in our decision to install a dedicated 1.5-T system at our institution.
The current study contributes to the growing literature on the role of high-field (1.5-T or 3.0-T) iMRI in neurosurgical procedures (1–4). It is now imperative to ask for what pathological conditions and presentations does iMRI provide true clinical benefit?
Justin F. Fraser
Philip H. Gutin
New York, New York
1. Hall WA, Liu H, Maxwell RE, Truwit CL: Influence of 1.5-Tesla intraoperative MR imaging on surgical decision making. Acta Neurochir Suppl
2. Hall WA, Truwit CL: Intraoperative MR-guided neurosurgery. J Magn Reson Imaging
3. Hirschberg H, Samset E, Hol PK, Tillung T, Lote K: Impact of intraoperative MRI on the surgical results for high-grade gliomas. Minim Invasive Neurosurg
4. Sutherland GR, Kaibara T, Louw DF: Intraoperative MR at 1.5 Tesla—Experience and future directions. Acta Neurochir Suppl
Jankovski et al. provide an illustrative report on a multidisciplinary installation of a 3-T iMRI unit. They present valuable information concerning issues such as shielding, hygiene, safety, and electromagnetic compatibility, information that might prove to be of special value to anyone planning to introduce high-field MRI to the OR. This report indicates how meticulously such a setup must be prepared to satisfy the needs of all involved parties. In addition, all of the demonstrated efforts to contain costs are imperative to convince the hospital administration to invest in this technology.
As we observe a trend toward the use of 3.0-T MRI in routine diagnostic imaging, the question as to what extent intraoperative imaging acquired with either a 0.15- or 3.0-T system influences patient outcome remains unanswered. Seemingly, neurosurgery follows the common trend of technology evolving faster than its benefits can be evaluated. As this trend will not be reversed, combined efforts will have to be made to rate the clinical impact of image-guided surgery in general and of iMRI in particular.
It is obvious that higher field strength enhances the spectrum of intraoperative imaging capabilities, extending to areas such as functional MRI, spectroscopy, real-time navigation, and vascular imaging. For the near future, however, intraoperative 3.0-T installations will remain restricted to a very few centers; thus, lower-field units provide an accessible and also sensible solution for most institutions, at least in the near future. In our experience, the image quality delivered by a PoleStar N20 (0.15-T) system must be rated as good in far more cases than have been stated by Jankovski et al. Current data demonstrate that this system generates reliable information about the extent of tumor resection in high-grade as well as in low-grade gliomas.
Still, to promote iMRI and to eventually establish it as a standard of care, the promise of the technology must first be proven. The newly founded International Intraoperative Imaging Society will certainly help to reach that goal.
Jankovski et al. report on implementation and preliminary experience with a 3-T iMRI suite. A total of 21 patients underwent 3-T perioperative MRI, and in 9 patients actual iMRI with the head only partially closed was performed. Although several low-field and an increasing number of high-field iMRI setups have been established in recent years, only a few reports on ultra-high-field iMRI, which has been pioneered at a few sites, are available. This article adds highly valuable information on the feasibility of ultra-high-field iMRI.
It will be of special interest to see the actual differences in the application of 1.5-T iMRI. What are the potential benefits of going to ultra-high-field iMRI? Is it better resolution of the border zone in gliomas or better visualization of detailed structures in the cavernous sinus during transsphenoidal surgery of pituitary tumors? Or is it the potential for much improved functional imaging capabilities with a better resolution in functional MRI and diffusion tensor imaging or even improved magnetic resonance spectroscopy (which still has not been shown to actually work intraoperatively)?
A 3-T imaging system offers a better signal-to-noise ratio, which could also be used to further shorten intraoperative imaging time. With our intraoperative 1.5-T system applying parallel imaging techniques, we need only about 15 minutes with an additional maximum of 5 minutes in total for moving the patient from the operating position into the scanner and back again until surgery continues, even for complicated procedures. For the application of intraoperative imaging, a smooth integration into the surgical workflow is crucial. The reported iMRI examination times of 78 ± 20 minutes with a maximum of 129 minutes are really unacceptable in a routine setting. Extreme cases of a waiting time of half an hour for the scanner to be available for intraoperative use may be attributed to the implementation phase and might be avoided in the future.
Sophisticated integration of multimodal navigation with intraoperative imaging including automatic registration of the pre- and intraoperative image data to update the navigation system during surgery, allowing highly reliable multimodal navigation throughout the whole procedure without any brain shift handicaps, seems to be mandatory to achieve extended resections with low rates of neurological deficits.
Jankovski et al. should be encouraged to investigate the potential benefits of intraoperative 3-T scanning especially compared with standard 1.5-T scanning: it will be very interesting to see how the full potential of a 3-T MRI system can be actually used in an operating room environment with advanced anatomic, structural, functional, and metabolic imaging in the future.
In this study from the University of Louvain in Brussels, Jankovski et al. presented a novel solution to the desire for iMRI. Instead of pursuing one of the commercially available systems and planning how to convince hospital administration to commit well over 1 million dollars to iMRI, they built one themselves—and not just any iMRI system, but one with a 3-T magnet. This project was only feasible as a shared resource system, in which the scanner is used for diagnostic imaging when it is not in use during surgery. Three years passed between conception to the first operation being done. The various technical and engineering issues relating (among others) to anesthesia, the sterile field, acoustic noise, air flow and pressure, and the magnetic environment were methodically solved. Of note is the fact that construction was done as part of an OR remodeling project; hence, use of iMRI does not require the surgeon to move to the radiology department or to a separate and unfamiliar OR complex.
The Louvain approach separates the OR completely from the magnet. Thus, surgeons can operate without any constraints in terms of equipment (although an MRI-compatible headholder is needed). To achieve intraoperative imaging, patients are moved on a mobile table on tracks to the MRI room entrance. From there they are transferred using an MRI-compatible plastic tabletop onto the scanner. Most of the time, the scanner is used for diagnostic imaging and thereby becomes a revenue source for the hospital.
Still, despite the ingenuity of the solution and the careful engineering described, the benefit of this elaborate technology remains unclear. In their early experience, Jankovski et al. reported using iMRI in 21 patients, with only two scans being acquired per operation. It may be that the 77 minutes average time it took to complete a patient transfer and image lessened the surgeons' zeal for acquiring intraoperative images. In only 3 of 20 patients undergoing tumor resections (18 patients) or epileptic foci (2 patients) did imaging lead to further resection. In 3 other patients residual tumor was identified on imaging but deemed not safely resectable. The authors also point out the vagaries of image interpretation in 5 patients; the neuroradiologists changed their reports upon their postoperative “second look” at the scans.
Surgery in the 3-T environment is not risk free. Two patients sustained second-degree burns, presumably because of looped wires that heated up during placement in the magnet, and as the authors noted “the behavior of the medical staff must be adapted to prevent any objects from becoming dangerous projectiles.”
Dr. Raftopoulos's group has shown ingenuity and stamina in bringing this project to fruition. No doubt with further experience they will smooth over some of the technical obstacles to routine actual intraoperative imaging. We should withhold any skepticism regarding the need for such high-field iMRI, and the need for elaborate patient transfers, pending follow-up reports and objective analyses from these and other centers.
Manhasset, New York
In this article, Jankovski et al. reported on the development and initial clinical use of an interventional MRI system based on a 3-T magnet. The patient is moved to the magnet for imaging according to their configuration, which allows the technology to be shared between diagnostics and surgery. This is important, given the cost of purchasing, installing, and maintaining an interventional MRI system. At the time of article submission, the system had been used to evaluate surgical procedures in 21 patients. The main limitation of the reported configuration remains the need to move the patient with its associated risk and impact on OR time. This limitation may have in part restricted the case mix to predominantly central nervous system neoplasia.
Although preliminary and somewhat limited, the application of surgical planning iMRI may have allowed diagnostic images to be updated and craniotomy placement to be optimized. Jankovski et al. demonstrated the application of interdissection imaging for additional tumor resection in three patients. Perhaps related to moving the patient, technical difficulties were encountered in 10 imaging studies, which were interpreted by Jankovksi et al. as a learning curve.
More than 10 years ago, we developed an iMRI system that was based on a moveable 1.5-Tesla magnet (1–3). The philosophy of moving the magnet allowed a patient-focused environment and enabled the sharing of technology between diagnostics and surgery. The system has been used as an adjunct to neurosurgery in some 1000 procedures that have included the full spectrum of neurosurgical procedures. Similar to Jankovski et al., we performed surgical planning imaging after the induction of anesthesia and patient placement to optimize craniotomy placement. In 20% of 202 patients with low-grade gliomas and 30% of 66 patients with pituitary adenomas, interdissection images showed unsuspected residual tumor. Among 139 patients treated surgically for temporal lobe epilepsy, interdissection images showed unsuspected residual tissue in 19%. Because of radiofrequency coil design, spinal applications have been restricted to the upper cervical spine for which 23% of 36 patients showed unsuspected residual tumor degenerative tissue.
In 89 patients with vascular pathological lesions, the benefit was restricted mainly to craniotomy placement. Quality assurance imaging performed after wound closure but before reversal from anesthesia confirmed that surgical objectives had been achieved and defined acute complications. In only one patient with a glioma was the surgical wound reopened for removal of hematoma.
Limitations related to iMRI include increased OR time (on average 90 minutes in our series), disruption of the surgical rhythm for image acquisition, and cost. Despite these, we have found that the benefits offset these challenges. Similar to Jankovski et al., we are upgrading our present iMRI system to a 3-Tesla system with a 70-cm working aperture. The large working aperture and the ability to acquire images rapidly and drape the magnet rather than the patient should collectively decrease imaging time. To offset cost, it will become increasingly necessary for intraoperative or interventional MRI technologies to be shared between health care departments.
Garnette R. Sutherland
1. Kaibara T, Saunders JK, Sutherland GR: Advances in mobile intraoperative magnetic resonance imaging. Neurosurgery
2. Sutherland GR, Kaibara T, Louw D, Hoult DI, Tomanek B, Saunders J: A mobile high-field magnetic resonance system for neurosurgery. J Neurosurg
3. Sutherland GR, Latour I, Greer AD: Integrating an image-guided robot with intraoperative MRI: A review of the design and construction of neuroArm. IEEE Eng Med Biol Mag