5 Functional Neuroimaging I: fMRI and Resting State fMRI

Abstract

Functional magnetic resonance imaging (fMRI) is a useful strategy for presurgical planning and operative guidance in neurosurgery. This chapter will review what fMRI measures, applications of presurgical fMRI for localizing individual-specific functional anatomy, and methodological issues that should be considered when interpreting functional neuroimaging data in individual cases. The focus will be on planning for tissue resection in epilepsy and tumor surgery. The current applications and potential uses of functional neuroimaging for guiding deep-brain stimulation implantation in movement disorders, chronic pain and psychiatric disorders will also be described.

In the common “task-based” fMRI approach, brain activity is measured while a patient performs tasks that involve sensorimotor, language and memory functions to localize eloquent cortex. Recent developments suggest that “resting-state” fMRI, in which spontaneous brain activity is measured while a patient is in a task-free state, is a feasible, and potentially more useful, alternative for presurgical mapping of a broader range of functional regions. In both approaches, due to low signal-to-noise, data analysis issues, and the correlational nature of the approach, fMRI data in individual cases must be interpreted with caution. Ongoing research efforts are leading to increased sensitivity and specificity of functional neuroimaging, which may enable more effective and broader applications for personalized functional neurosurgery in the future.

The human brain is composed of local regions that have specialized functions in sensation, movement, and cognition. Yet, a survey map of anatomical landmarks is often unable to provide accurate inference about the functional importance of a given brain region, and the potential impact of injury to it. Thus, noninvasive functional neuroimaging has emerged as an attractive strategy for surgical guidance.

Functional neuroimaging methods, including functional magnetic resonance imaging (fMRI) and positron emission tomography (PET), involve indirect measures of neural activity, relying on changes in cerebral blood flow or metabolic activity that reflect the electrophysiological signaling of neuronal populations. ¹ Blood-oxygen-level dependent (BOLD) fMRI, which is sensitive to the magnetic properties of deoxyhemoglobin in blood, is the most popular approach. In BOLD fMRI, with standard scanners of 1.5–3 Tesla strength, whole-brain activity is sampled every 2–3 seconds in spatial units of cubic voxels that are 2–4 mm³. New advances enable higher resolution. ² Thus, in gray matter locations, voxel-level activity may reflect the pooled activity of hundreds of thousands of neurons. The BOLD response is said to be ‘sluggish’ in that it begins several seconds after an increase in neuronal activity, and plateaus after 6–12 seconds. ¹ The BOLD signal is partly contaminated with non-neuronal ‘noise.’ Examples of such contamination include scanner-related and head-motion artifacts, respiratory and cardiac signals. Care must then be taken in data acquisition and its analysis.

fMRI is predominantly a research tool. Presurgical mapping is its most common clinical application. It is most often used for epilepsy surgery and tumor resections, but its role is expanding to involve planning for implanted electrodes in the treatment of movement disorders, chronic pain, and psychiatric disorders. The goal of presurgical fMRI, is to locate eloquent cortex such as in primary sensory, motor and language-dominant regions. Another application that has been used with limited success is fMRI for localization of epileptic tissue (see also Chapter 20 for discussion of EEG and imaging). ³ The relative effectiveness of fMRI may vary depending on the specific goal and case.

In presurgical fMRI, a “task-based” approach is used in which patients may perform such tasks as finger-tapping, over multiple time blocks that are interspersed with task-free periods. These tasks have not been standardized. Importantly, task-based fMRI is a correlational approach. For example, increased motor cortex activation during tongue movement does not indicate that the activated region is necessary and sufficient for tongue movement. Thus, since the earliest applications of presurgical fMRI, results have been verified using the effects of electrical cortical stimulation following neurosurgical implantation of electrode contacts. ⁴ For example, does stimulation of the BOLD-active region cause tongue movement? While there is often good overlap between fMRI results and electrical stimulation mapping of early sensory areas, the correspondence is not always straightforward. The fMRI results can also be compared to those from the more invasive intracarotid amobarbital procedure, or “Wada” test, in which a barbiturate is injected into the right or left internal carotid artery to inactivate one hemisphere. Results of task-based fMRI often show correspondence with the Wada test for mapping language- and memory-related functions. Results are more reliable for language than for memory testing. Currently, the American Academy of Neurology recommends that presurgical task-based fMRI for certain epilepsy subtypes be considered for lateralizing language, and in limited cases, memory functions, in place of the Wada test. ⁵

Task-based fMRI requires time for instruction and training, expertise to conduct, and patient cooperation and compliance. Furthermore, the benefit of functionally localizing sensorimotor areas is often unclear, as they can be broadly identified anatomically with structural MRI. A distinct approach, known as “resting-state” (rs-fMRI, has been developed as a clinically feasible and potentially more effective approach for mapping a broader range of functional regions. ⁶ In rs-fMRI, the patient is instructed to not think about anything in particular for several minutes. Spontaneous brain activity is analyzed with correlations of BOLD time series among remote regions of the brain. Regions that have synchronized activity are said to exhibit “functional connectivity,” denoting integration into a network. Each brain region can be classified according to the network(s) to which it belongs. Research in healthy adults has revealed that segregated networks throughout the brain, including those related to sensorimotor, language, and memory processes, are readily identifiable and are reproducible in rs-fMRI. ⁷ Moreover, presurgical rs-fMRI functional connectivity of sensorimotor regions has been shown to correspond with the results of intracranial cortical stimulation better than with task-based activation. ⁸

The rs-fMRI method is attractive both because of its demonstrated effectiveness and because of the ease of implementation. It requires little active participation from the patient. Additionally, theoretical concepts from basic neuroscience support the use of rs-fMRI. Converging evidence suggests that spontaneous correlated activity, when considered over at least several minutes, reflects the ‘intrinsic’ organization of the brain that is partly constrained by anatomical connectivity and that is predictive of functional co-activation during tasks. ⁹ Spontaneous activity is continuous and thus its study is not limited to the resting-state context. Indeed, task-based fMRI lasting several minutes can be analyzed with both task-based activation and spontaneous functional connectivity approaches. In a new development, the combination of these two analyses outperformed individual analysis of identifying sensorimotor regions that correspond with the output of direct cortical stimulation. ¹⁰

However, several issues in functional connectivity analysis remain unresolved. Compared to task-based fMRI, the analysis is more error prone from non-neural noise factors such as head motion. Special care is needed in data preprocessing, although no consensus has yet been reached on best practices. ¹¹ Moreover, the ideal scan duration remains unknown. Scans longer than six minutes or multiple scans are known to produce more reliable results but may not always be clinically feasible. ¹² In both task-based and rs-fMRI, the data processing strategies and statistical thresholds used are variable. Standards for general fMRI data analysis are beginning to emerge, ¹³ and some such practices could be adopted clinically.

Although less common in practice, task-based and functional connectivity approaches to functional neuroimaging have been applied to psychiatric disorders. ¹⁴ In treatment-resistant major depressive disorder, PET and fMRI have been used to identify mood circuitry to target with deep brain stimulation. ¹⁵ In addition, there is evidence that intrinsic rs-fMRI networks have relevance for personalized targeting of regions related to cognitive and affective dysfunctions. For example, intracranial stimulation of a mid-cingulate cortex region that was individually mapped to the rs-fMRI “salience network” caused a stereotyped cognitive and affective experience that could be described as a “will to persevere.” ¹⁶

Future such research may lead to a broader range of applications for personalized functional neurosurgery.

Related posts:

1 History of Innovation in Stereotaxy/Functional Neurosurgery 8 Microelectrode Recordings in Deep Brain Stimulation Surgery for Movement Disorders 22 Epilepsy: Extra-Temporal Surgery with Invasive Monitoring 29 Neurosurgery Treatment for Anorexia Nervosa 36 Ablative Procedures for Trigeminal Neuralgia 39 Neuroprosthetics

Stay updated, free articles. Join our Telegram channel

Join