Transform Based Method for Classification of Various Meningioma Subtype Images

Abstract

A brain tumor is an abnormal mass of tissue in which some cells grow and multiply at a rapid speed. This growth of a tumor occupies space within the skull, interferes with normal brain activity and causes damage to the brain tissue and nerves by increasing pressure in the brain, by shifting the brain or pushing against the skull. Identification of the tumor requires a neurological examination, scanning of the brain followed by analysis of the brain tissue and biopsy. The overall process is time consuming. Once the tumor has been diagnosed the next challenge involves determining the type of brain tumor. Of these Meningioma tumor accounts for 27% of the entire brain tumor. So the next challenging task is classification between the subtypes of Meningioma tumor viz. Fibroblastic, Meningiothelial, Transactional and Psammomatous. In this paper, we present a wavelet based technique for discriminating between four different subtypes of Meningioma. Because the overall process, right from the diagnosis of the tumor to the classification of the Meningioma subtype is time consuming, may also cause error and mainly dependent on the experts convenience, hence computer based technique helps in classification of Meningioma subtypes.

Keywords: Brain tumor, Meningioma, Fibroblastic, Meningiothelial, Transactional and Psammomatous, Wavelet based technique.