| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Radiology, University Hospitals of Cleveland/Case Western
Reserve University, Cleveland, OH 44106.
2 Present address: Department of Radiology, Johns Hopkins University, Outpatient
Center, 601 N Caroline St., Rm. 4210, Baltimore, MD 21287.
3 Department of Neurosurgery, University Hospitals of Cleveland/Case Western
Reserve University, Cleveland, OH.
4 Present address: New Mexico Neurosurgery, Albuquerque, NM.
5 Siemens Medical Engineering Group, Erlangen, Germany.
Received September 21, 2006;
accepted after revision June 4, 2007.
The University Hospitals of Cleveland/Case Western Reserve University
Interventional MR Program is supported in part through research collaborations
with Siemens Medical Solutions and Radionics. This project was also supported
through grants from the Prentiss Foundation, Whitaker Foundation, and American
Cancer Society.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. One hundred twenty-two patients (65 males, 57 females; age range, 6–77 years; mean age, 43.8 years) underwent 130 neurosurgical and ENT procedures (106 craniotomies, 17 transsphenoidal pituitary resections, three biopsies, three intracranial cyst aspirations or injections, and one skull base resection) in a specially designed surgical MRI suite equipped with a 0.2-T imager and a prototype rotating, tiltable surgical table. The intraoperative MR sequences included free induction with steady-state precession (fast imaging with steady-state precession [FISP]), steady-state free precession T2-weighted, reverse fast imaging with steady-state free precession (PSIF), FLASH, spin-echo T1-weighted, turbo spin-echo (TSE) T2-weighted, and TSE FLAIR. Each case was analyzed for the number of imaging sessions, duration of each session, total imaging time during surgery, and impact of imaging information on procedure.
RESULTS. Each patient underwent between one and five intraor postoperative imaging sessions. Imaging times were 1.7 seconds–8 minutes 31 seconds per sequence. The mean total imaging time was 35 minutes 17 seconds per surgical procedure. Imaging was continuous during biopsy and cyst aspiration procedures and averaged 200.67 and 54.66 minutes, respectively. Additional surgical resection based on intraoperative imaging findings was performed in 72.8% of the cases.
CONCLUSION. Intraoperative low-field MRI provides valuable information for surgical decision making that is predominantly related to detection of residual tumor and the exclusion of complications. The benefits of this technology surpass the time cost associated with its implementation when using proper imaging strategies.
Keywords: brain neoplasms • gliomas • interventional MRI • intraoperative MRI • MR technique • neuroradiology • neurosurgery
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The room was constructed with two radiofrequency doors: one leads directly into the sterile corridor connecting all operating suites and the other leads into the MR control room, which has a door that leads into an adjacent nonsterile corridor (Fig. 1). The surgical flooring is color-coded to allow ready identification of the 20-, 0.5-, and 0.15-mT field lines (Fig. 2A, 2B, 2C, 2D). A 0.2-T imager (Open Viva, Siemens Medical Solutions) was installed along with a prototype surgical table. This table can be positioned within the magnet isocenter and can then be smoothly rotated to a 120° angle from the imager, placing the surgical field for procedures on the patient's head, chest, abdomen, and spine outside the 20-mT field line (Fig. 2A, 2B, 2C, 2D). By reversing the patient position, pelvic and lower extremity procedures are also possible. Unlike earlier intraoperative MRI, in which the patient remained in a horizontal position throughout the procedure, this table allows a wide range of height adjustments and tilting to both Trendelenburg and reverse Trendelenburg positions. The latter position was used almost universally during the surgical procedures to reduce brain edema. The Trendelenburg position was not typically used during this series.
|
|
|
|
|
Several interventional accessories were also installed in the surgical MR suite. A radiofrequency-shielded LCD monitor with a resolution of 1,024 x 1,280 pixels was installed to allow in-room image viewing (Fig. 2A). A frameless stereotaxy system was created in which optical tracking of an imaging-guidance probe was accomplished with a 3D optical digitizer system (Polaris System, Northern Digital). This digitizer uses two cylindric lens video sensors to localize infrared light-reflecting markers mounted on a handheld probe or surgical instrument and provides an infrared light source to allow reflection. Unlike earlier forms of integrated digitizer–MR systems [8], no wire connection to the handheld instrument is necessary, facilitating sterilization and use of the instruments. The frameless stereotaxy system calculates the probe tip's position and orientation in camera coordinates from the spatial location of the reflective markers. This position and orientation are translated into scan offset and image plane rotation in the coordinate system of the MR imager and are provided to the MR system for subsequent image acquisition either in a continuous mode or in a single or multishot fashion.
No historical image data set is used with the digitizer system; rather, images related to the probe position are prospectively acquired. Scanning planes are defined in two different modes: first, conventional axial, sagittal, and coronal planes relative to the patient in which the image center is at the probe tip, which we refer to as the "patient view"; and, second, planes orthogonal, parallel, or both orthogonal and parallel to the probe centered at the probe tip, which we refer to as the "probe view."
Imaging was performed using a circularly polarized head coil, a large multipurpose solenoidal coil, or a prototype sterilizable solenoidal coil with surgical pin head fixation. The latter coil was developed and supplied by the Heidelberg Neurosurgical Research Group (Fig. 2D).
Surgical Equipment
Standard surgical instruments were tested to determine the distance from
the magnet at which torque was measurable, and this distance was noted for
each instrument. Titanium versions of all instruments used in the main
magnetic field, including retractors, specula, and curettes, were obtained or
instruments were left in place during imaging. Standard operating microscopes,
electrocautery instruments, a cortical stimulator, and fiberoptic headlamps
were used (Fig. 2C). Endoscopy,
when used, was performed with an MR-compatible scope, light source, camera,
and LCD display. Brain biopsies were performed using the Navigus System
(Image-Guided Neurologics).
Subjects and Procedures
One hundred twenty-two patients (65 males and 57 females; age range,
6–77 years; mean age ± SD, 43.8 ± 20.2 years) underwent
130 neurosurgical or ear, nose, and throat surgical procedures in this
surgical MRI suite. Written informed consent was obtained from all patients
under a protocol approved by the institutional review board for human
investigation. Surgical procedures included craniotomy (n = 106),
transsphenoidal pituitary resection (n = 17), brain needle biopsy
(n = 3), intracranial cyst aspiration or injection (n = 3),
and skull base resection (n = 1). Craniotomies were performed for
resection of tumor (n = 81), treatment of seizures (n = 21),
removal of cranial abscess (n = 1), and arteriovenous malformation
with hematoma (n =3).
Intraoperative Imaging Protocols
Intraoperative pulse sequences included free induction with steady-state
precession, fast imaging with steady-state precession (FISP), steady-state
free precession T2-weighted, reverse fast imaging with steady-state free
precession (PSIF), and both 3D and 2D T1-weighted FLASH gradient-echo
sequences, along with T1-weighted spin-echo, T2-weighted turbo spin-echo
(TSE), and TSE FLAIR. For each sequence, between 3 and 19, 3- to 5-mm sections
were obtained with a 20- to 25-cm field of view, and 128–256 x 256
matrix. Imaging times varied from 1.7 seconds to 8 minutes 31 seconds per
sequence. The specific parameters and imaging times for each sequence are
outlined in Table 1.
|
For craniotomies, each imaging session typically included contrast-enhanced T1-weighted FLASH images in all three orthogonal planes, along with T2-weighted images in a subset of cases. Baseline scans were obtained after craniotomy and before the commencement of surgical resection. For transsphenoidal surgery, each imaging session included coronal or sagittal (or both) T2-weighted images and 3D coronal T1-weighted images; coronal and sagittal contrast-enhanced T1-weighted images were also obtained during the final imaging set. The specific sequences used for other pathologic conditions were more variable depending on the patient's anatomy and the imaging characteristics of each lesion on the initial intra- or preoperative images.
For biopsy needle or catheter insertion guidance or cyst aspiration or injection, a continuous series of FISP or PSIF images was acquired, reconstructed, and automatically displayed on the in-room monitor with a frame rate of between 1.7 and 5 seconds per image.
A neuroradiologist was present during key portions of the imaging sessions to oversee the imaging protocols and to interpret the imaging findings. A senior MRI technologist was also present to operate the scanner throughout table repositioning and imaging.
Data Evaluation
Each case was analyzed for the number of imaging sessions during the
surgical procedure, the duration of each imaging session, the impact of the
imaging information on the surgical procedure, and the total time spent on
imaging during surgery. The impact on surgical procedure was based on whether
additional tumor resection was performed because of imaging findings. Note was
also made when the surgeon decided no further resection was necessary on the
basis of the images. The duration of the imaging sessions was separately
calculated for intraoperative (more resection after imaging) and postoperative
(last imaging session of case) sessions. This division was made because the
final session of each case, once the decision was made that further resection
was unnecessary, included additional investigational sequences for academic
purposes rather than clinical need in certain types of procedures.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
|
|
|
|
|
|
|
|
|
|
The time necessary to drape the patient's head, reposition the coil (if necessary), position the patient's head at the magnet isocenter, and tune the system ranged from 30 to 90 seconds. Movement of the table back into the operating position typically took less than 1 minute. The rotating movement of the table allowed the operating microscope and other large pieces of equipment to remain in place during table positioning, and surgery could continue immediately after table positioning was complete.
The mean total time spent on imaging per case for all cases except patients undergoing biopsy or cyst aspiration was 35 minutes 17 seconds per surgical procedure. Imaging for brain biopsy and cyst aspiration or injection was continuous during the procedures and averaged 200.67 and 54.66 minutes of the total imaging time, respectively (mean time, 127.67 minutes). Evaluation of the impact of imaging on surgery revealed that additional surgery was performed due to the presence of residual resectable tumor seen on imaging in 72.8% of the cases.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The first system introduced for exclusive intraoperative or interventional MRI was the mid-field (0.5-T) double-donut system (Signa SP, GE Healthcare) [3, 9–12]. Reports from investigators at various institutions subsequently followed that outlined their experiences with intraoperative imaging using different magnet designs and field strengths, ranging from a portable 0.12-T MRI unit to a full closed-bore 1.5-T system [3, 4, 6, 13–16]. A review of the literature highlights an existing trade-off among the various available systems between a host of factors, including the initial cost of the system, access to the patient, instrument artifacts, acquisition times, and imaging capabilities.
As with other interventional MRI setups, the ultimate intraoperative neurosurgical imaging environment would allow unrestricted patient access while offering real-time high-quality images with minimal instrument artifacts. In reality, maximizing access to patients through open-magnet designs, along with facilitating procedures in the vicinity of conventional surgical equipment and monitoring devices, typically entails a low-field approach.
A recognized limitation of intraoperative imaging at low field strengths is the increased imaging time required to achieve sufficient image quality for surgical decisions to be reliably made. In the current investigation, we report the clinical results of 130 neurosurgical and ENT procedures in 122 consecutive patients performed under low-field MRI in a university setting, evaluate the additional operative time associated with intraoperative low-field MRI guidance, and discuss methods to minimize this time without significantly compromising image quality.
In this series, the use of intraoperative MRI allowed neurosurgeons to make the necessary dynamic adjustments for the brain shift associated with dural opening and CSF drainage. The incorporation of intraoperative imaging during surgical procedures also facilitated more accurate intraoperative lesion localization that was based on real-time information rather than on historical data collected from preoperative imaging and stereotactic localization procedures. Other attributes of intraoperative imaging that are significant advantages over conventional surgical approaches include the ability to guide and thereby to optimize surgical navigation; the ability to assess the completeness of tumor resection while conserving the maximum amount of normal brain tissue; and the ability to intraoperatively detect procedure-related complications such as hematoma.
Corroborating the results presented in prior reports [2, 17–20], the results of our study emphasize the value of intraoperative MRI for the complete and safe resection of diffusely infiltrative brain neoplasms, which typically have margins that are poorly delineated on visual inspection of the surgical field despite the use of high-power magnification and the illumination offered by operative microscopes.
Many of the neoplasms suitable for imaging-guided surgery in our series were brain gliomas. Among these, patients with low-grade gliomas constituted the major beneficiary sector owing to the particular difficulty of distinguishing tumor margins from adjacent normal white matter and to the curative effect of complete resection of these less aggressive neoplasms when compared with high-grade gliomas. Claus et al. [21] recently reported significantly increased 1-, 2-, and 5-year survival rates for patients who underwent surgical resections of low-grade gliomas using intraoperative MRI guidance.
The prototype tilting and rotating surgical table coupled with the open C-arm magnet design allowed convenient switching of the table between the imaging and operating positions in a time-efficient manner. Furthermore, adjustments to the table height and acquiring reverse Trendelenburg position were readily available as needed. This integrated system also facilitated entirely unrestricted access to the patient so that surgeons were able to implement their standard surgical approaches and use conventional instruments while freely surrounding the surgical field outside the 20-mT magnetic field line.
An experience similar to ours was described by Rubino et al. [22], who reported the proper function of standard operating equipment in the 0.5- to 10-mT zone of an open 0.2-T MRI system during 22 craniotomies and 16 brain biopsy procedures. Repositioning a patient protruding from a cylindric system back to the isocenter of the magnet may be easier as described by Rubino et al. than in the system we describe. Patient repositioning and system tuning did not, however, require more than 30–90 seconds in our series. In addition, surgeons did not have to modify their approaches or instruments to fit a high-field environment or a nonadjustable table.
The low magnetic field strength associated with the open-system design implies an inherently limited signal-to-noise ratio (SNR). Nevertheless, the intraoperative images generated on the open 0.2-T system were of sufficient quality to impact the surgical decision making in a high proportion (72.8%) of the cases in our series, predominantly through the detection of resectable residual tumor. This finding substantiates the results of Steinmeier et al. [6], who reported the successful use of low-field intraoperative MRI to detect residual tumors and resection-associated hemorrhage and to delineate pituitary neoplasms in a cohort of 55 patients.
The construction of high-quality, inherently low-noise receiver chains has significantly enhanced the image quality of low-field systems by improving the SNR on images obtained by these low-field resistive or permanent magnet systems; however, an increase in imaging acquisition time compared with high-field systems is still required due to additional signal averaging and an increase in the number of phase steps.
To minimize the additional acquisition time while providing neurosurgeons with sufficiently informative images, several technical factors can be modified to increase the SNR. For example, the use of quadrature receiver coils can produce an increase in SNR of up to a factor of 2. In addition, modifying pulse sequence parameters—such as increasing the field of view, decreasing the matrix size, increasing the slice thickness, and using low-bandwidth techniques—helps increase the acquired signal without the need for additional imaging time. Furthermore, implementing alternative image reconstruction techniques such as keyhole imaging, wavelet encoding, and singular value decomposition has been associated with reduced imaging acquisition times [23–32].
Applying these strategies, we were able to maintain the total additional operative time resulting from imaging within a mean of 35 minutes 17 seconds per surgical procedure despite the multiple intermittent imaging sessions (range, 1–5 sessions; mean ± SD = 1.95 ± 0.77) performed per case. This was not considered prohibitively long by our neurosurgeons in the context of the types of surgical procedures performed.
Although the discussed strategies were largely successful in providing satisfactory images in reasonable time frames, the low-magnetic-field environment is, in a way, a limitation to the current investigation. High-field scanners, in addition to offering higher SNR and spatial and temporal resolutions, enjoy a full armamentarium of functional capabilities—such as MR angiography, diffusion-weighted imaging (DWI), MR spectroscopy, and brain activation and perfusion studies—that can potentially refine the intraoperative information available to the surgeon and thereby enhance the safety of tumor resections [33–35]. Nimsky and colleagues [36] reported a large series of successful intraoperative high-field MRI (1.5-T) examinations with integration of functional data. They have also recently shown the feasibility of intraoperative visualization of major pyramidal tracts using DWI-based fiber tracking [37]. The most recent work to date was reported by Pamir et al. [38] and involved intraoperative imaging on a 3-T scanner built in an interconnected room contiguous with the operation theater.
Another limitation of intraoperative MRI that currently applies to both low- and high-field approaches is the necessity to move the patient back and forth between the scanning and operating positions. This setup provides intermittent "snapshots" on which to track the progress of surgery rather than offering a real-time environment for procedure guidance. One advantage of the open low-field system, as shown in the current study, is the feasibility of continuous interactive guidance during biopsy and aspiration procedures without the need to remove the patient from the scanner.
Finally, objective assessment of intraoperative surgical guidance remains a difficult task to achieve. A potential limitation of the current investigation is that the herein reported significant additional resection rate (72.8%) may be subject to a nonmeasurable bias resulting from the ease of intraoperative imaging. Surgeons typically switched to the intraoperative imaging when they believed that resection was complete. However, the convenient logistics of the intraoperative MRI setup and the ability of surgeons to obtain decent-quality MR images in a relatively timely manner have sometimes led them to opt for checking the operative progress with imaging in situations when they would otherwise proceed with additional resection. This practice did, however, help enhance the safety of surgical resections, a basic aim of acquiring intraoperative MRI scans.
Our results generally corroborate the data in the literature regarding the impact of intraoperative MRI on the immediate surgical progress, with most of the reported rates of additional resection based on intraoperative imaging ranging from 50% to 92% [29, 39–45]. Nimsky et al. [36] did, however, report a much lower rate of 27.5% despite using high-field intraoperative MRI. Again, variations in the local practice pattern of neurosurgeons likely contribute to this discrepancy.
In conclusion, the results of the current investigation show the value of intraoperative MRI in a large cohort of consecutive neurosurgical and ENT patients with various pathologic conditions. Our results support the notion that the benefits of this technology surpass the time cost associated with its implementation. The abilities to detect residual disease during the excision of infiltrating malignant brain neoplasms and to exclude complications intraoperatively are the core merits of this approach over standard surgical approaches. Low-field intraoperative MRI systems are associated with lower SNR and with limited spatial and temporal resolutions and lack functional capabilities compared with their high-field counterparts. Conversely, the improved patient access and monitoring, minimal instrument artifacts, and feasibility of interactive guidance during needle-based procedures are advantages of the low-field approach. New state-of-the-art high-field open-configuration scanners that have recently become available will likely provide myriad advantages of both systems in the near future.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
