Extraction of Brain Tumor in Coronal MRI Sliced Images by Implementing Active Contour and Different Segmentation Techniques

Abstract

Many segmentation methods were proposed to process medical images for diagnosis purposes. In this work, two stages were achieved to extract tumor region of three MRI sliced brain images of coronal orientation. Active contour algorithm was implemented as first step to extract brain matter of T2W_FLAIR modality MRI images for the first time to avoid interference of tumor gray values with skull. The second stage involved implementing four segmentation techniques to extract tumor regions. First technique was utilizing contrast adjustment process after presenting analysis study of this operation to select an adequate gray level range to be stretched. Second method is an adaptive technique, which is contour based method, to isolate and extract tumor regions. The third and fourth methods were K-Means and FCM. The surface and relative surface area of the extracted tumor regions were calculated. The results of extracting brain matter proved high quality performance of active contour algorithm with its adaptive initialization element. The adequate gray level range that deduced from the proposed analysis of contrast adjusting process is [0.6 0.7]. In addition, the results of the proposed contour based technique present good isolation and extraction of tumor regions. The four implemented techniques are adequate to extract tumor regions correctly without overlapping with skull as declared from the results. This work was implemented by utilizing the facilities of Mat lab programing system.