Data Mining Applied to fMRI Clinical Data

Functional Magnetic Resonance Imaging (fMRI) of children with epilepsy is of special interest in patients with prospective surgery. Doctors need to locate the areas in the brain associated with basic functions, such as language and motor areas in order to optimize the surgical outcome and minimize any deficits as a result of the surgery. At the Center for Advanced Technology and Education a large language fMRI dataset has been collected from such patients, and the fMRI datasets have been analyzed and classified.  However, the clinical data, such as genre or age, has not been utilized. In this study we will take employ techniques from Data Mining and try to correlate the clinical data to the results obtained by the classifications. Collected data from five different hospitals is stored in a relational database. With the aid of a unique data mining specific fMRI patterns are extracted and then related to those patterns obtained from previous classifications which serve as historical basis. The results show that patients versus control populations affected significantly results of group separation; handedness also affected significantly the results of group separation (with right handedness considered typical and the rest as atypical). The patient population and atypical handedness have more cases classified into the right dominant category. Excluding the right dominant category, the same tests were carried out with the patient vs. control populations and handedness and the results show no difference.