Advanced clinical MRI for better outcome in epilepsy surgery. Focusing on fMRI and prediction of verbal memory decline.

Abstract
Aim: The aim of the thesis was to evaluate the use of advanced MRI technology to improve results of epilepsy surgery,
with focus on language and memory functions.
Methods: In paper I, 25 patients with drug-resistant epilepsy were retrospectively included in the study for having been
referred to high resolution 3T MRI with and without surface coils. The surface coils were placed over the suspected
epileptogenic zone. The efficacy of the coils was assessed and graded in relation to their placement. In papers II, III and
IV, a functional MRI (fMRI) paradigm, including both a verbal encoding task and a visuospatial task, was designed and
implemented. The medial temporal lobe (MTL) for memory and the anterior language area were studied. In paper III, a
standard word generation fMRI paradigm was also included. In these three studies, the test was performed in 15 healthy
right handed subjects (paper II), 6 patients with drug-resistant temporal lobe epilepsy (TLE) with mixed handedness and
10 controls (paper III) and 14 TLE patients eligible for resective surgery (paper IV), respectively. A bootstrap algorithm
was used to calculate lateralization indices (LI) and LI-curves. In paper IV, a clinical risk assessment score was created from
collected clinical data. Additional value from fMRI LI was correlated to post-surgical memory decline.
Results: Surface coil 3T MRI did not contribute to detection of previously undiagnosed lesions. In 20% of patients, 3T
MRI, compared with previous 1.0-1.5T MRI, provided new information about cortical lesions. The fMRI paradigm
visualized memory-related activity in the MTL and provided information regarding language processes. LI and LI-curves
for memory at group level were consistent with previous studies, but a variety of activation effects were found at the
individual level. LI-curves added complementary information for individual subjects with uncharacteristic results. The
verbal encoding task provided information on verbal memory, which had equal lateralization as language in right-handed
subjects, but not always in left-handed subjects and TLE patients. Bilateral fMRI language representation was seen in 2
right TLE patients who later suffered verbal memory decline post-operatively.
Conclusion: High resolution 3T MRI is valuable for lesion detection, but surface coils do not provide further crucial
information. The fMRI paradigm activates memory and language areas that can be studied at an individual level using LIcurves.
Analysis of language patterns seems to be important for prediction of memory outcome in both left and right TLE.
fMRI indices may identify an unexpected high risk for post-operative verbal memory decline in right TLE patients.