Characterization of Brain Tumor Using Haralick Parameter and Tumor Volume Calculation

CT images have excellent bonny details with the ease of availability. But due to less contrast and details it is less studied. CT images of 5 tumor identified patient were procured. Then this study is divided into three parts. (1) Characterization of tumor using texture analysis. (2) Segmentation of the tumor and (3) volume calculation of the tumor. Preprocessing is important in order to remove the noise and further analysis of image. It is done via contrast enhancement and using median filter the noise is removed. In order to determine the image characteristic we applied texture analysis including Homogeneity, Correlation, Contrast, and Energy. Paired t-test using SPSS software is applied to find the significance of data of tumorous and non-tumorous image. Segmentation and extraction of tumor is performed via Watershed and Fuzzy c-means algorithms. Both the algorithms were evaluated for correctness and completeness. The watershed shows superiority over fuzzy c-means as it lacks robustness. Lastly, volume of the brain tumor is evaluated using MATLAB ® software and compared with the manual results.