This thesis deals with a segmentation of brain tissues from MRI image data and its implementation in MATLAB.
Segmentation problematic is described with attention to formulating segmentation as optimization problem and segmentation of given images with different metaheuristic algorithm consequently. This approach was chosen due to information from last specialized publications, where it was accentuated for its fast computational speed and universality. This thesis tries to prove this statement with segmentation of brain images with brain tumours that have different types, number, stage of illness and phase of therapy.
For more information check this paper https://dspace.vutbr.cz/xmlui/handle/11012/138153
or whole thesis on my university page https://www.vutbr.cz/studenti/zav-prace/detail/110519?zp_id=110519
Achieved result:
Manual:
The program start runnig using script main.m. It is required to set some variables as path to images in BRATS folder: variable nac_obrazy. Algorithm is able to load only images in .mha format. You can select the the image from BRATS folder that you want t segment, you can choose the metaheuristic algorithm (FA, SSO or hybrid FASSO) and the segmentation algorithm (2D or 3D). Default setting is to image HG0015 using 3D segmentation and FASSO algorithm.
Output of the program is the JACCARD score and visualization of segmented image.
As we talk about metaheuristic algorithm, user can select the most basic optimalization algorithm paraneters too..
Example of flowchart for 3D segmentation
Date 25.05.2018 class="center"