Probabilistic mutual information based extraction of malignant brain tumors in MR images

Extraction of the malignant tumor region from the brain magnetic resonance (MR) image is a critical task. As the soft tissues of the brain neoplasm has a lot of variation thus the proper extraction of the malignant tumor and segmenting the affected part from brain MR image is the major part of concern. Malignant brain tumors like Central Neuro Cytoma (CNC), Glioblastoma Multiforme (GBM), Gliomas, Intra Ventricular Malignant Mass and Metastasis are concerned as one of the critical brain tumors of medical science therefore the extraction and appropriate segmentation of such tumor from the brain MR image in their early phase of generation is required. In the past innumerable approaches had applied on brain MR imaging system to figure out the proper abnormality region inside the brain image. The literature gives an idea about various clustering and segmentation approaches that help to identify the abnormal region inside imaging systems based on predefined threshold values. While the selection of proper threshold is again a mind blundering. In this paper a new image segmentation algorithm is proposed which clusters the maximum possible abnormality region based on the probabilistic mutual information. Proposed algorithm is free from any initial selection of threshold value and gives significant results in extraction of the malignant tumor region inside brain MR image.