Functional magnetic resonance imaging (fMRI) enhances the understanding of neuroanatomy and functions of the brain and is becoming an accepted brain-mapping tool for clinicians, researchers, and basic scientists alike. A noninvasive procedure with no known risks, fMRI has an ever-growing list of clinical applications, including presurgical mapping of motor, language, and memory functions. fMRI benefits patients and allows neurosurgeons to be aware of, and to navigate, the precise location of patient-specific eloquent cortices and structural anomalies from a tumor. Optimizing preoperative fMRI requires tailoring the fMRI paradigm to the patient’s clinical situation and understanding the pitfalls of fMRI interpretation.
Functional magnetic resonance imaging (fMRI) has been used to enhance the understanding of neuroanatomy and functions of the brain and is becoming an accepted brain-mapping tool for clinicians, researchers, and basic scientists alike. A noninvasive procedure with no known risks, fMRI has an ever-growing list of clinical applications, including presurgical mapping of motor, language, and memory functions. By superimposing fMRI data onto high-resolution anatomic magnetic resonance images, the location of eloquent cortices relative to a lesion can be determined. This information gives the neurosurgeon the opportunity to make more informed decisions regarding the approach to the tumor and the necessity for invasive mapping procedures. fMRI data can also guide intraoperative mapping techniques, such as direct cortical stimulation and somatosensory evoked potentials. This article reviews the applicability of fMRI to clinical neurosurgical practice, describes the optimization of paradigm design and delivery, and illustrates artifacts and other clinically relevant pitfalls of fMRI.
Clinical importance
The use of fMRI in the clinical setting benefits patients insomuch as it allows neurosurgeons to be aware of and to navigate the precise location of patient-specific eloquent cortices and any structural anomalies that may have developed from a tumor. This anatomic/functional preview facilitates the creation of more effective patient-specific treatment plans. Before the use of fMRI, the preoperative location of eloquent cortices and their relationship to the lesion was determined based on the historical localization of function and performing intraoperative mapping using direct cortical stimulation. However, direct cortical stimulation has several drawbacks. First, this procedure requires a craniotomy. Hence, an adjustment of the operative plan and possible expected complications may ensue, should intraoperative mapping reveal a relationship of the eloquent cortex to the lesion than is different from what was assumed using historical data. Further, in the authors’ experience, patients occasionally have difficulty cooperating with task performance during cortical stimulation, especially when attempting to map higher function, such as language. As a result, operations can be terminated early because of inadequate intraoperative mapping results that might have been foreseen and avoided with a reliable preoperative mapping technique such as fMRI. A lack of preoperative functional information may also cause a lesion to be unnecessarily deemed inoperable. In addition, mapping is restricted to the exposed surface of the brain and as a result, the cortex in the deep sulci cannot be mapped. Lastly, invasive mapping procedures, such as cortical stimulation, do not aid the neurosurgeon in presurgical risk assessment and patient counseling.
Preoperative fMRI, however, by defining eloquent motor, language, and/or memory areas, allows a surgeon optimize perioperative planning to maximize tumor resection while minimizing damage to surrounding areas ( Fig. 1 ). Planning a surgical resection relies heavily on the anatomic location of a tumor relative to the eloquent cortices. However, the locations of the functional cortices vary with individuals and may be affected by the lesion itself and/or associated cortical reorganization and thus may not match the predicted anatomic locations. fMRI allows for the clarification of anatomy in specific patients and detection of any irregularities. These details may become even more important when the eloquent cortices are infiltrated by a tumor.
In addition, fMRI data can help a surgeon decide either for or against resection by providing information about tumors in high-risk locations (involving Broca’s area, Wernicke’s area, and so on) without invasive mapping (in some cases), sparing the patient awake anesthesia and associated potential complications. If a surgeon decides to perform surgery, fMRI helps the surgeon to choose the best trajectory for resection (eg, taking a more posterior approach to bypass an important functional area). Other benefits of using fMRI before surgery include gathering data to guide intraoperative direct cortical stimulation (or other mapping procedures) and decreasing the operation time. In patients for whom invasive mapping is not an option, fMRI provides a safe alternative, affording an opportunity for better comparable care.
fMRI can also mitigate the consequences of failed intraoperative cortical stimulation. Such failures may result from the limited time available for various testing paradigms or the inability of the patient to cooperate with the paradigms while conscious and undergoing cortical stimulation with an exposed cariotomy. Optimum sedation for an open craniotomy may prove difficult to achieve, and patients may not be able to follow instructions. If awake mapping fails, previously acquired fMRI data can offer information to the surgeon.
Coregistration to Neuronavigational Systems
The application of fMRI to guide neurosurgical procedures has been greatly facilitated by the ability to register fMRI with high-resolution MRI data and to display this coregistered information in the operating room in real time. Coregistration of fMRI onto high-resolution anatomic images allows for better visualization of lesions when compared with the quality of images from low-resolution fMRI (T2*-weighted images). Typical in-plane resolution for functional images is lower than that of routine magnetic resonance images (for example, 64 × 64 matrix vs 256 × 256 matrix). After analyzing the fMRI data and determining active voxels in an fMR image, these low-resolution data are coregistered onto high-resolution images to reveal the precise location in the brain in which these signal changes occurred. The coregistered images are then downloaded to the neurosurgical navigational system in the operating room, which allows the neurosurgeon to view the relationship of the lesion to the adjacent eloquent cortex in real time intraoperatively. This whole process can be accomplished using commercially available software packages. Using the coregistered data, the neurosurgeon can also envision the 3-dimensional relationship between the lesion and adjacent eloquent cortices during the resection itself, allowing for more precise patient-specific functional localization.
Verification of fMRI Results
The accuracy of fMRI results, in general, has been validated by a variety of techniques. For instance, magnetoencephalography (MEG), which is based on electric activity, has shown excellent concordance. When compared with results of the sodium amobarbital procedure (also known as the Wada test), fMRI has shown good concordance with determinations of hemispheric dominance. Studies have also shown that data collected by both intraoperative electrocorticography and direct cortical stimulation are comparable to those collected by fMRI for motor mapping. In a study by Jack and colleagues, 100% of subjects had concordant fMRI and direct cortical stimulation results regarding sensory/motor areas.
Integration with Other Techniques
fMRI used in conjunction with electroencephalography (EEG) or magnetoencephalography (MEG) may allow for the tracking of neural networks and a better understanding of the relationships, or connections, between brain areas used to perform a task. Whereas both EEG and MEG are best in the temporal domain (timing resolution <1 millisecond), fMRI has superior spatial resolution (as small as 1.5 mm). In this way, the techniques can be complimentary. Also, it is significant that fMRI and EEG can be performed simultaneously, allowing for more extensive data acquisition related to one-time events, such as seizures. The combination of methods, therefore, lends complimentary information that allows for a more complete and clinically useful picture of brain function.
Paradigm selection and patient preparation
Neuroanatomic Review
A brief review of basic anatomy of the motor and sensory system facilitates the discussion of paradigm selection. The motor and sensory systems both have a topographic organization; in other words, motor and sensory functions are mapped to specific locations on the cortex. The foot and leg are represented along the interhemispheric fissure, the hand lateral to that of the foot and leg, and the tongue and face lateral to that of the hand ( Fig. 2 ).
Voluntary movement is performed through a network of different motor areas: the primary motor cortex (M1), the supplementary motor area (SMA), the lateral premotor cortex, and the superior parietal lobules. M1 is involved in actually performing movement, whereas the SMA is involved in motor planning and organization. SMA, located in the superior frontal gyrus, consists of 2 parts: the rostral pre-SMA and the caudal SMA proper. The pre-SMA functions in cognitive tasks and language, whereas the SMA proper, which is anatomically closer to primary motor and sensory areas, is involved in sensory and motor planning and word articulation.
The role of the motor strip in language processes, although not completely understood, should be acknowledged in preoperative fMRI/planning because it has been shown that silent speech paradigms, most commonly used for language mapping, may reveal only negligible inferior motor area activity, primarily showing only activation of the dominant inferior frontal gyrus (frontal speech areas). Vocalized speech fMRI paradigms may help better estimate the network responsible for speech production (both silent and vocalized speech production), allowing patients to retain normal speech capacity after surgery. This result is particularly true given that the surgeon measures speech localization in terms of speech disruption of vocalized speech rather than speech production. This finding again reinforces the complementarity of fMRI and direct cortical stimulation. fMRI identifies speech production, whereas direct cortical stimulation delineates speech arrest, which are clearly related but not equivalent.
Paradigm Design
With the most commonly requested fMRI examinations being language and sensory/motor related, it is important to consider the aforementioned neuroanatomy (specifically the anatomy of the motor homunculus) in light of the lesion’s location in selecting paradigms. Paradigms should be designed according to certain guidelines. First, paradigms should help neurosurgeons and neurologists offer the best care to patients by providing physicians with critical information that they can use to consider the benefits and risks of neurosurgery and other treatments. Second, paradigms should be selected based on the position of the tumor and the surrounding functional areas (eg, language, motor and sensory, memory). Lastly, paradigms should be chosen such that the task can be performed by the patient adequately while in the MRI scanner, taking into account the patient’s age, neurologic deficits, and general medical history.
Common paradigm designs include block and event-related paradigms. To acquire meaningful results, a patient typically performs the fMRI task (the “ON” state) 3 to 10 times, with breaks, or “OFF” states, between each ON state. When performing a block paradigm, patients alternate between ON and OFF periods of equal or unequal duration (nonperiodic task delivery). To identify the hand motor area, for instance, a block paradigm might consist of alternating finger-tapping and resting periods, each lasting 20 seconds. Nonperiodic task delivery often proves effective in decreasing noise from the scanner, heartbeat, and breathing.
In an event-related paradigm, a patient performs a single event (such as swallowing or clenching a fist), which lasts much shorter than the ON period in the block design. The OFF period lasts the same duration as in the block paradigm. This type of paradigm is used to investigate the neuronal or hemodynamic response to a specific single event. In event-related paradigms, statistical power for the events being measured is compromised; hence, these designs typically require more images to regain the statistical power obtained with a block paradigm. However, the advantage of event-related paradigms is that the hemodynamic parameters such as the time to peak, full width half maximum, and return to baseline can be estimated in a way that is not possible with block designs. These more advanced parameters allow one to study complicated neurologic phenomena, such as cortical reorganization.
Regardless of the paradigm used, it is important that it be designed to minimize any type of unwanted head or body motion to achieve ideal results. A toe paradigm, for example, should be performed with no ankle motion, and tongue paradigms should be performed with a closed mouth. Another method of limiting motion is using pillows under the knees or behind the neck.
Sensory and Motor Mapping
Sensory and motor mapping often entails simple paradigms that use the previously described ON/OFF method. Mapping of the motor cortex involves finger, toe, and/or tongue movement paradigms that help identify the relative locations of the lesion and regions of the motor homunculus ; paradigms chosen should determine the position of the lesion with regard to the different areas of the motor strip. For lesions close to the midline, leg or foot paradigms are used (because the foot and leg motor areas are represented along the interhemispheric fissure).
Sensory paradigms can identify the sensory cortex or can be used to determine the location of the motor cortex in paretic patients. A sensory paradigm may include brushing, squeezing, or touching the patient’s foot or hand. Because there is significant reciprocity between the motor and sensory gyri, such a sensory paradigm often elicits M1 activation as well ( Fig. 3 ). Even without motor homunculus fMRI activation during a sensory paradigm, the location of the motor gyrus can often be deduced from the sensory information. This can serve as a useful trick in a neurologically compromised patient.
Patient Preparation and Scanning Optimization
Another crucial step in obtaining optimal results is patient preparation to ensure proper paradigm delivery and patient compliance. Many difficulties that may occur during an fMRI scan can be avoided through a thorough preprocedure screening and paradigm explanation. This procedure is especially true for more complex paradigms and in neurologically compromised patients. Patients should arrive early to receive an explanation of the fMRI procedure and the timing of the paradigms. They should then practice paradigms that they will be asked to perform because this will help to foresee possible complications. Paradigms may also be modified to better suit the patient’s neurologic limitations. It is difficult for some neurologically impaired patients to follow instructions and actually perform the paradigms; hence, patient progress and performance should be continuously monitored to achieve meaningful and helpful results.
It is important to properly position the patient in the scanner and to properly select the scanning plane. For motor paradigms, the scanning plane is usually parallel to the axial plane, intersecting the anterior and posterior commissures (AC-PC line). If the scanning plane is not properly oblique to the AC-PC line, the motor gyrus will be artifactually pushed back on the image, making anatomic judgments (even with functional information) difficult.
Paradigm selection and patient preparation
Neuroanatomic Review
A brief review of basic anatomy of the motor and sensory system facilitates the discussion of paradigm selection. The motor and sensory systems both have a topographic organization; in other words, motor and sensory functions are mapped to specific locations on the cortex. The foot and leg are represented along the interhemispheric fissure, the hand lateral to that of the foot and leg, and the tongue and face lateral to that of the hand ( Fig. 2 ).
Voluntary movement is performed through a network of different motor areas: the primary motor cortex (M1), the supplementary motor area (SMA), the lateral premotor cortex, and the superior parietal lobules. M1 is involved in actually performing movement, whereas the SMA is involved in motor planning and organization. SMA, located in the superior frontal gyrus, consists of 2 parts: the rostral pre-SMA and the caudal SMA proper. The pre-SMA functions in cognitive tasks and language, whereas the SMA proper, which is anatomically closer to primary motor and sensory areas, is involved in sensory and motor planning and word articulation.
The role of the motor strip in language processes, although not completely understood, should be acknowledged in preoperative fMRI/planning because it has been shown that silent speech paradigms, most commonly used for language mapping, may reveal only negligible inferior motor area activity, primarily showing only activation of the dominant inferior frontal gyrus (frontal speech areas). Vocalized speech fMRI paradigms may help better estimate the network responsible for speech production (both silent and vocalized speech production), allowing patients to retain normal speech capacity after surgery. This result is particularly true given that the surgeon measures speech localization in terms of speech disruption of vocalized speech rather than speech production. This finding again reinforces the complementarity of fMRI and direct cortical stimulation. fMRI identifies speech production, whereas direct cortical stimulation delineates speech arrest, which are clearly related but not equivalent.
Paradigm Design
With the most commonly requested fMRI examinations being language and sensory/motor related, it is important to consider the aforementioned neuroanatomy (specifically the anatomy of the motor homunculus) in light of the lesion’s location in selecting paradigms. Paradigms should be designed according to certain guidelines. First, paradigms should help neurosurgeons and neurologists offer the best care to patients by providing physicians with critical information that they can use to consider the benefits and risks of neurosurgery and other treatments. Second, paradigms should be selected based on the position of the tumor and the surrounding functional areas (eg, language, motor and sensory, memory). Lastly, paradigms should be chosen such that the task can be performed by the patient adequately while in the MRI scanner, taking into account the patient’s age, neurologic deficits, and general medical history.
Common paradigm designs include block and event-related paradigms. To acquire meaningful results, a patient typically performs the fMRI task (the “ON” state) 3 to 10 times, with breaks, or “OFF” states, between each ON state. When performing a block paradigm, patients alternate between ON and OFF periods of equal or unequal duration (nonperiodic task delivery). To identify the hand motor area, for instance, a block paradigm might consist of alternating finger-tapping and resting periods, each lasting 20 seconds. Nonperiodic task delivery often proves effective in decreasing noise from the scanner, heartbeat, and breathing.
In an event-related paradigm, a patient performs a single event (such as swallowing or clenching a fist), which lasts much shorter than the ON period in the block design. The OFF period lasts the same duration as in the block paradigm. This type of paradigm is used to investigate the neuronal or hemodynamic response to a specific single event. In event-related paradigms, statistical power for the events being measured is compromised; hence, these designs typically require more images to regain the statistical power obtained with a block paradigm. However, the advantage of event-related paradigms is that the hemodynamic parameters such as the time to peak, full width half maximum, and return to baseline can be estimated in a way that is not possible with block designs. These more advanced parameters allow one to study complicated neurologic phenomena, such as cortical reorganization.
Regardless of the paradigm used, it is important that it be designed to minimize any type of unwanted head or body motion to achieve ideal results. A toe paradigm, for example, should be performed with no ankle motion, and tongue paradigms should be performed with a closed mouth. Another method of limiting motion is using pillows under the knees or behind the neck.
Sensory and Motor Mapping
Sensory and motor mapping often entails simple paradigms that use the previously described ON/OFF method. Mapping of the motor cortex involves finger, toe, and/or tongue movement paradigms that help identify the relative locations of the lesion and regions of the motor homunculus ; paradigms chosen should determine the position of the lesion with regard to the different areas of the motor strip. For lesions close to the midline, leg or foot paradigms are used (because the foot and leg motor areas are represented along the interhemispheric fissure).
Sensory paradigms can identify the sensory cortex or can be used to determine the location of the motor cortex in paretic patients. A sensory paradigm may include brushing, squeezing, or touching the patient’s foot or hand. Because there is significant reciprocity between the motor and sensory gyri, such a sensory paradigm often elicits M1 activation as well ( Fig. 3 ). Even without motor homunculus fMRI activation during a sensory paradigm, the location of the motor gyrus can often be deduced from the sensory information. This can serve as a useful trick in a neurologically compromised patient.
Related posts:
Special Surgical Considerations for Functional Brain Mapping
PET and SPECT in Brain Tumors and Epilepsy
An Introduction to Diffusion Tensor Image Analysis
Multimodal Image Registration for Preoperative Planning and Image-Guided Neurosurgical Procedures
Identification of Neural Targets for the Treatment of Psychiatric Disorders: The Role of Functional Neuroimaging
Development of a Clinical Functional Magnetic Resonance Imaging Service
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
