Medical Data Analysis Method For Epilepsy

摘要

Applying data mining techniques on medical databases which contain un-structured and semi-structured data is a challenging task. It is not only due to the complexity of such databases but also due to the characteristics of the medical domain. This thesis describes how multiple layers of data mining techniques have been applied to a Human Brain Image Database system. It starts with data preparation which paves the way for conventional data analysis techniques to be applied to the data. A similarity based patient retrieval tool has been designed and developed to assist in treatment planning and outcome estimation for epileptic patients. Finally connected scatter-plot visualization tool has been designed and implemented in order to assist the medical experts to see the relationship between attributes and visually compare a new patientu27s similarity scores against patients that have been previously operated on in the hospital.