Preoperative Functional Localization

Principles and Limitations of Functional Magnetic Resonance Imaging

Functional MRI (fMRI) as a technique for language mapping became a favored method due to many advantages; noninvasive, greater spatial and temporal resolution than IAT, and the facility to repeat testing and employ a greater stimulus set.

The role of cognitive fMRI in presurgical assessment is to lateralize, localize, and predict the risk for language and memory outcome after epilepsy surgery. Although there is great variability in the cognitive paradigms, scanner parameters, and statistics employed in functional imaging studies, a consensus is beginning to emerge with published practical guidelines.

Developed in the 1990s, fMRI relies on the blood oxygen level–dependent (BOLD) changes associated with physiologic and pathologic processes in the brain. ^, The BOLD effect captures neurovascular coupling as a surrogate marker of synaptic activity where there is increased glucose and oxygen utilization. With a particular cognitive task, there is an increase in cerebral blood flow to the brain regions involved with a flux from deoxygenated to oxygenated blood. Deoxyhemoglobin is a paramagnetic molecule that causes local field inhomogeneity, while oxyhemoglobin is diamagnetic and has little effect on MR images. This differing magnetic properties forms the basis of the “contrast” in fMRI.

Echo planar imaging and spiral MRI are examples of MR sequences used to detect BOLD changes as they allow an entire image to be acquired in a single excitation to increase the temporal resolution of functional imaging. These images are, however, prone to susceptibility artefact, particularly within the temporal lobes. This is a result of the field inhomogeneities introduced by the different magnetic properties of bone, tissue, and air. Solutions to this include “unwarping” and shimming, acquiring scans with opposite-phase encoding polarities, and reducing the field of view of imaging depending on the cognitive task employed. Given the test-retest variability of fMRI activations, longitudinal studies investigating plasticity of cognitive networks after surgery must include healthy controls scanned across the same time points, controlling for between-subject and between-group variance.

Language Lateralization

There is general concordance of language fMRI lateralization with amytal testing. The American Academy of Neurology (AAN) practice guidelines suggest that language fMRI may be used to lateralize language function in temporal and extratemporal epilepsy, and language fMRI is now used in most surgical centers.

Language function is carried out by a network. ^, At least seven regions have been shown to be relevant in lexico-semantic tasks using fMRI. ^, These include the Broca area (inferior frontal gyrus); posterior middle frontal gyrus; supplementary motor area; angular gyrus; inferior and superior Wernicke area (middle to anterior superior temporal gyrus [STG], posterior STG, and supramarginal gyrus); basal temporal language areas; and the anterior temporal cortex that are involved in auditory naming. Localization and preservation of these sites is important to prevent postoperative decline in language function. The tasks commonly used to lateralize language include semantic decision, verb generation, verbal fluency, auditory response naming, word generation, and sentence reading. ^{, ,} In addition, the American Society of Functional Neuroradiology (ASFNR) hosts presentation files and parameters of the common paradigms on their website.

A lateralization index (LI) of fMRI activations at the whole-brain level or within a specified region of interest gives a unitary measure of how lateralized brain activations are. Word production tasks such as verbal fluency and verb generation activate frontal regions more than posterior temporal or basal regions. ^{, ,} Activation of temporal lobes are appropriate for understanding the functional anatomy of naming and receptive tasks. ^, Multiple tasks may be used to gain a comprehensive picture of hemispheric dominance.

Localization and Prediction of Language Outcome After Surgery

For surgery, a critical question is whether the area to be resected represents eloquent cortex. Activation within an area of resection is predictive of naming decline after surgery. This applies to right as well as left temporal resections. In one recent study, 3 of 21 right TLE patients were right-lateralized for picture naming within the temporal lobes, and all 3 showed significant decline in naming function after right anterior temporal lobe resection, exemplifying the need for individual localization paradigms in presurgical assessment.

Auditory and picture-naming fMRI tasks show reliable activation of the posterior and basal temporal lobe. Activation in the left posterior inferior temporal gyrus on auditory naming and in the left fusiform gyrus on picture naming relates to greater naming decline after surgery. ^{, ,} When a posterior temporal resection in a language-dominant hemisphere is considered, other tasks that involve posterior language areas (posterior STG, supramarginal gyrus) should be considered, including the phonologic retrieval stage of speech production such as repetition, naming, and reading.

Asymmetry of language fMRI using a verbal fluency task within a frontal region of interest appears to be a sensitive predictor of language outcome after temporal lobe surgery, but it has a low specificity. Using an auditory description decision-making task, the fMRI LI within a frontotemporal mask is 100% sensitive and 75% specific compared with direct cortical stimulation in predicting language outcome after epilepsy surgery. In a picture-naming task, a left lateralized LI within a large temporal lobe mask predicts naming decline with 100% sensitivity and 92% specificity after dominant temporal lobe resection ( Fig. 88.1 ).

Memory Processes

Short-term memory refers to the temporary storage and manipulation of information, while long-term memory involves the storage and retrieval of information. Explicit (declarative) and implicit (nondeclarative) memory systems make up long-term memory processes. Critical steps of episodic memory include memory encoding, memory storage, and memory retrieval.

Lateralization of Memory Functions

People with TLE may have long-term memory deficits affecting semantic, autobiographical, and episodic processes. Neuropsychometry has shown deficits in verbal memory in left TLE. Visual memory deficits have been shown to be less consistently lateralizing for right TLE. Earlier memory encoding fMRI studies showed concordance of asymmetry of mediotemporal lobe (MTL) activations with intracarotid amobarbital memory LIs. Material-specific paradigms help investigate reorganization of memory functions caused by the epileptic process. Whole-brain fMRI studies have shown the importance of extratemporal episodic memory activations to memory encoding. Both temporal and extratemporal activations have been shown to be material-specific with reorganization to involve homologous contralesional and ipsilesional brain regions. Reorganization of memory functions to the posterior temporal lobe including the posterior hippocampus has been shown to be related to an earlier age at onset of epilepsy. Longer duration of epilepsy and higher seizure burden were associated with contralesional extratemporal activation. The AAN practice guidelines suggest that memory fMRI may be useful to lateralize memory functions in people with mesial temporal lobe epilepsy, but further studies are required for other epilepsy syndromes.