Chapter 2 Surgical Navigation with Intraoperative Imaging
Special Operating Room Concepts
For high-grade gliomas, in 1994 Albert reported that post-operative imaging showed tumor remnants in 77% of patients who were presumed to have undergone gross total resection.1 In 2006,2 Stummer et al. published a multicenter randomized study, which, as a byproduct, showed residual tumors in 64% of the patients undergoing conventional microsurgical tumor resection (only patients with high-grade gliomas were included, which was deemed—by imaging criteria—to be fully resectable). With the importance of the extent of resection for high-1,2 as well as low-grade gliomas,3,4 these findings emphasize the need for improvement.
In neurovascular surgery the routine use of intraoperative angiography has been advocated to avoid undetected residual disease.5,6 In spinal surgery, the significant percentage of misplaced screws could be reduced from 10%, but still occurs with approximately 5%, even with modern navigation techniques.7 These findings underscored the desire to complement advanced preoperative evaluation with intraoperative quality control. Thus various surgical groups proceeded to integrate imaging into their procedures.
The earliest attempts were made with ultrasound (US) and computed tomography (CT). The immediate impact on surgical procedures was small, due to limited resolution (US and CT) and cumbersome integration into the operating room (CT). Another avenue opened with the introduction of image-guided neuronavigation (IGN) systems.8,9 These systems allowed the transfer of increasingly refined presurgical image information into the operating theater to guide surgical procedures. However, intraoperative changes (“brain shift”) critically limited their application accuracy.10,11 The concept of intraoperative imaging resurfaced. With magnetic resonance imaging (MRI) becoming the method of choice for the imaging of the central nervous system, pioneering efforts to introduce this modality into surgery provided proof of the concept.12–14 These initial experiences with intraoperative MRI (iMRI)15–17 ignited diversification into a variety of approaches.
Computer-Assisted, Image-Guided Neuronavigation
The major link between imaging and integration of this information into surgery is provided by navigation systems. Diagnostic computer-based image-analysis and three-dimensional (3D) modeling facilitated the spatial definition of complex pathologic processes. The desire to use this information directly in the surgical field led to the introduction of IGN systems in the mid-1980s8,9 and their commercial availability in the early 1990s. These systems provided the surgeon with a tool that allowed the transfer of presurgical image information in an intuitive and interactive fashion into the surgical field (see Chapter 3 for more detail on neuronavigation).
Meanwhile, the technology has proceeded from being a novelty to an established asset for neurosurgical procedures. Questions of prior consideration, that is, application accuracy and integration of instruments, were overcome. However, the major shortcoming was the dependence on preoperative image data. Since intraoperative changes (e.g., CSF drainage, tumor resection, sagging of the cortex, swelling of underlying tissue, summarized as “brain shift”), accumulate throughout surgery, preoperative data become invalidated.10,11,18 This has particular influence on glioma surgery. While enabling precise approach planning and localization, resection control is generally beyond the capacity of these systems, since they cannot account for intraoperative changes. Intraoperative imaging resolved this issue directly. It enables continued use of these systems with newly acquired accurate data.
A different avenue investigates mathematical models to compensate for brain shift. Various algorithms can characterize and calculate deformation matrixes.10,11,19 Various brain shift patterns were identified. A multimodal approach appears potentially useful, which uses intraoperative “sparse” US data20–22 to calculate a deformation matrix, which is then used to elastically deform preoperative MRI images. Albeit all these efforts advances were meager and the only option to provide precise updated navigation remains the integration of intraoperative images.
Intraoperative Imaging
Intraoperative Fluoroscopy
Operating theaters for stereotactic neurosurgery had built-in biplane x-ray to eliminate parallax artifacts in imaging of electrode placement. With the limited scope of this application, these ORs remained rare and have largely been replaced by standard fluoroscopy, or more recently intraoperative MRI.23,24
In instrumented spinal surgery, fluoroscopy is used as an online imaging modality for planning and verifying screw positioning. Combinations with navigation systems have been propagated. Intraoperative angiography has been employed by major vascular centers for quality insurance in aneurysm and AVM surgery.5,6
Intraoperative Ultrasound
Intraoperative US (IoUS) was one of the first to be employed as an intraoperative imaging modality in neurosurgery.25 With subsequent new generations, image quality improved and miniaturization of the hand-pieces enhanced applicability. Advantages are the dynamic, surgeon-driven, on-line character of the information.26 Particularly in vascular surgery, the flow-related analysis of duplex sonography provides additional flexibility. Further major developments were the introduction of spatially accurate 3D ultrasound,27 of contrast agents28 and the integration of US into navigation systems.26,29–31 In particular, the last aspect provided the means for easier interpretation of the images, which generally demands experience.
For the last 20 years, IoUS has been regarded as the most promising system for online information acquisition in neurosurgery. Still, these systems remain limited in their distribution. Potential reasons may be the unfamiliarity with the technique of ultrasound and its limitations in tissue differentiation,32 differing from the most widely distributed primary diagnostic modality of MRI.33
Major indications are circumscribed lesions, such as metastasis, cavernomas, vascular pathologies, and for spinal intradural lesions. With its integration into conventional navigation systems and in combination with iMRI34 the unfamiliarity with this modality might potentially be overcome.
Intraoperative Computed Tomography
Shalit and Lunsford first reported the integration of a stationary CT into OR.35,36 The next generation of CTs was mobile, permitting shared application in the OR and the ICU. However, image quality and radiation exposure limited the application and further implementation of this modality. Further advances in CT- and OR-table technology and integration with navigation systems have led to a reappraisal.
Modern CT-OR (Fig. 2-1) solutions use a rail system to move the CT between a parking position and the patient for scanning,37 which provides full access to the patient. In spine surgery, intraoperatively acquired images can be used to update navigation systems to provide additional image guidance for screw placement, as well as verification of correct positioning. For neurovascular surgery, intraoperative CT-angiography has the potential to provide information on obtained occlusion of vascular pathologies, but also with perfusion CT on potential vascular compromise.
Intraoperative Magnetic Resonance Imaging
Postoperative MRI remains the gold-standard for defining the extent of resection in neurooncology1,2 and pituitary lesions.38
The desire to employ the potential of MRI to monitor open neurosurgical procedures, as means to quality insurance, resection control, and complication detection led to the combination of MRI and surgery.13 Presently intraoperative MRI is used primarily for gliomas and pituitary lesions,38–40 but also for vascular41 and epilepsy surgery.42
The “twin operating theater”14,16 combined surgery and imaging (low-field, open 0.2 T MR system with a horizontal opening) by using two adjacent rooms. The patient was transferred between surgical and imaging site. Thus conventional OR equipment could be used without MR-safety or compatibility issues. To minimize the time for the transfer, this approach was modified by operating in the vicinity of the MRI, the “fringe field.”43,44
The open magnet design (“double doughnut”)12,13,17 aimed at a full integration of surgery and MRI. The vertical opening provided the surgeon with access to the patient. Surgical and imaging site were merged, a transfer was unnecessary. For practical reasons, surgery was discontinued during scanning. However, this design held the potential to provide real-time imaging, such as in biopsies, or through “continuous imaging” protocols.45 Furthermore, a navigation system was an integral part of the MRI. With a localizer, the surgeon controlled the scanning plane of the MRI interactively.46,47 Specially developed software for intraoperative navigation extended the functionality.48,49 This solution is closest to the symbiosis of surgery and imaging. However, by operating in a magnetic field, constraints in regards to technical equipment, in particular the microscope, the 56-cm gap for the surgeon, and the need for nonferromagnetic instruments, microneurosurgical standards were difficult to uphold.
These pioneering clinical experiences proved that the vision50 to bring MRI into the surgical surrounding could be realized. Biopsies as well as interstitial therapies could be blended with MRI into a novel procedure. However, it became evident, that the synthesis of open surgery and MRI into a comprehensive new method proved too complex. Either imaging potential, in comparison to preoperative high-field diagnostic scans, patient access or both, were restricted.
With emphasis on accessibility, a minimized, compact open MRI (0.12 T, 0.15 T) was introduced, which fit beneath the surgical table.51,52 To integrate high-field (1.5 T and higher) imaging, while providing ample patient access and only minimal influence on microneurosurgical instruments and techniques, surgical and imaging sites were separated. This can be achieved within an integrated OR-MR design (“dedicated”),15,40,53 or by arranging MR and OR into separate adjacent modules/rooms41,54–56 (“shared resources”).
A comprehensive classification, which encompasses present arrangements and accommodates potential future developments and expansions, cannot be based on variable characteristics such as field-strength and MR-design. Since the original concept was to merge surgery and imaging, it is reasonable to use work flow to distinguish among different installations. Specific issues for the integration of MRI into the surgical surrounding such as MR safety and compatibility of equipment, field strength, shielding, MR design,47,57,58 and imaging characteristics will be outlined before discussing various MR-OR integrations.
MR Safety, Compatibility, and Shielding
The introduction of a magnet into a surgical surrounding raises safety issues pertaining to interaction of the magnetic field and OR equipment.47,58–61 The magnet can exert a pull on ferromagnetic instruments. Generally the strength of the pull is related to field strength (and MR shielding) and distance to the MRI. The so-called 5-gauss line demarcates the inner area, in which the pull increases and the outer zone, in which ferromagnetic instruments can be safely used without being drawn into the MRI. In most MR-ORs, this demarcation is indicated on the floor. The immediate area around the 5-gauss line, which is within the magnetic field but still has no significant pull, is called the “fringe” field.
Shielding is necessary to prevent the interaction of the magnet with radio-frequency (RF) technology. Normally the entire room is shielded to prevent the magnet’s influence on electrical devices and vice versa. Alternatively, a specific shielding can be laced around the patient for scanning. While all nonessential electrical equipment can be turned off during scanning, or is primarily based outside the shielded room (e.g., the computer for image guided navigation), special anesthesia equipment is used to prevent RF noise (artifacts) in the images.
MR Design (“Open-Bore” and “Closed-Bore” Systems) and Field Strength
The static magnetic field of the MRI is generated within its bore. In open-bore (i.e., open-magnet) systems, the magnet is divided into two poles. The gap can be horizontal or vertical (“double doughnut”),59 resulting in different access to the patient.
Generally the open-magnet design has lower field strength than the “short-bore” closed systems. Higher field strength generally promotes acquisition speed (temporal resolution) as well as quality of the subsequently acquired images (spatial resolution). A wider range of image sequences is available (e.g., spectroscopy, DTI, fMRI, dynamic scanning).15,62,63 Furthermore, the homogeneity of the magnetic field increases, reducing geometric distortions. This issue is of major importance in low- and mid-field scanners. Phantom studies performed on the compact 0.12 T system provided acceptable application accuracy.52 However, studies in a stronger magnetic field (mid-field 0.5 T, open MR system) have shown that significant geometric distortions are present,64 which are machine- and patient-induced. These findings have to be considered when using non–high-field MR units (below 1 T) for resection control and updated neuronavigation.
Imaging
Which imaging to choose depends on the lesion’s imaging characteristics in diagnostic studies. Enhanced and nonenhanced T1WI, T2, and occasionally FLAIR answer most questions.39,40,53,54,62,65 For low-grade lesions, T2 and FLAIR images are the most appropriate.40,53,54 For enhancement, pre- and post-contrast T1 images are acquired.
Further sequences may potentially yield additional information,53 such as location of functional centers or fiber tracts. Both features can be extracted from intraoperative MRI, especially the latter.66
The intraoperative MRI is essentially a surgical tool. It is implemented to support surgical decision making. Thus the surgeon has to define his or her intention and the subsequent question, which primarily relates to the achieved extent of resection (residual tumor and its localization) and complication avoidance (distance to critical structures). It is essential that the surgeon acquires a good working knowledge of MRI to compile the individual imaging protocol and analyze the images according to surgical objectives.40,41,53,62
Practical challenges in interpreting intraoperative images largely pertain to nonspecific contrast enhancement (“spread enhancement”). The surgical result is described by “removed percent of contrast-enhancing lesion.” Since contrast enhancement merely reflects the local breakdown of the blood–brain barrier, it is unsurprising that contrast spreads into surrounding regions over time. While almost inconsequential in diagnostic imaging, acknowledging this phenomenon is of major importance for intraoperative MRI (iMRI) to avoid over-resection. Thus scans for the initial neuronavigation-assisted resection should be acquired prior to surgery. When imaging is for resection control, pre- and post-contrast T1 images and subtraction are compared to identify residual contrast enhancement. New sequences capturing the dynamic nature of neovascularized areas, in particular dynamic susceptibility contrast-weighted perfusion MRI (DSC-MRI), provide more accurate intraoperative information than conventional contrast-enhanced T1WI.67 Future development of specific contrast media may lead to a resolution of this problem.68,69
