Temporal lobe epilepsy is a neurological disease that affects millions worldwide, and can be effectively treated through surgical resection of portions of the anterior temporal lobe (ATL). Though effective for seizure control, this surgery occasionally produces language and verbal memory deficits. The goal of this work was to improve language-mapping techniques using fMRI in order to better inform the surgeon prior to ATL resection.
This was approached in three main experiments. First, an ATL deactivation task was developed. Since the ATL is hypothesized to be involved in semantic memory processing, mathematical stimuli were used, which are strongly attentionally engaging, but do not contain much semantic content. This was compared to previously used tasks and shown to be superior at producing ATL deactivation.
Second, an activation task was designed using story passages, differing from many previous studies that have used single word stimuli to activate the ATL. This task was combined with the math baseline task to produce maximal contrast in the ATL.
Finally, an improved thresholding method is proposed that increases the consistency of fMRI maps in individual patients, adapted for the requirements of surgical planning. While traditional methods of thresholding are effective for research questions, use of activation maps for surgical planning requires an approach that is less sensitive to individual variability in noise level. This new method was shown to be superior to current methods at predicting patient outcome.