Abstract
Within the health care industry, detection of brain tumor is a challenging issue. The cells of tumorous tissues grow uncontrollably; nonetheless, several techniques are available to detect those cells. This paper advances an innovative method for detecting tumorous cells using their thermal behavior. It advances a design to develop a fully automatic structure for brain tumor detection in Magnetic Resonance Imaging with good segmentation accuracy.
Regular metabolic activities of human beings are reflected in body temperature, indicating their physical condition. Brain tumor cells produce more heat than normal cells, and hence, it is more apt to use the bio-heat equation to exploit the thermal activities of the brain. The current scrutiny is performed in three steps. First, segmentation is applied to extract a region of interest using the Fast Bounding Box algorithm. Second, Cattaneo bio-heat equation is used to generate bio-heat distribution model of the living tissues on the region of interest. The generated bio-heat distribution model is used to analyze thermal diffusivity in living tissues numerically. Third, Extreme Learning Machine is then applied to the bio-heat distribution model to determine the size and location of tumorous tissues with more accuracy and efficiency. The proposed framework is implemented using MATLAB on the Multimodal BRATS database.