Detection and 3D modeling of brain tumors using image segmentation methods and volume rendering techniques

The abnormal and uncontrollable growth of cells inside the brain is called a brain tumor. Brain tumors are graded from the first degree to the fourth degree, from low to high depending on the risk they carry. Therefore, the sooner the tumor is detected and intervened, the more successful the treatment will be. Because the damaged brain tissue and its functional disorders are difficult to recover, it is important to detect the tumor early and remove it before a further spread. However, there is a risk of damaging healthy tissues during this operation. However, if you have more information about the tumor's location, size and how it spreads inside the brain, the success rate of the operation will be greatly increased and the risk will be minimized. Magnetic Resonance Imaging (MRI) is the most effective imaging method for the detection of the brain tumor. MRI images consist of 2D slices. With these images, the tumor can be detected, but no precise information about its shape and spread can be obtained from individual slices. However, it is possible to obtain more information by processing the MRI images. A 3D image can be obtained from 2D slices with volume rendering, but to obtain a 3D image of the brain tumor, first tumor tissue must be separated from the normal brain tissues using image segmentation algorithms. After a successful image segmentation, the 3D image can be obtained using volume rendering on processed images. In this thesis, with the developed software, MRI images are taken as input and they are automatically processed. After the process is finished and the detection is completed the 3D shape of the detected tumor is re-created and its features can be observed.