Segmentation of Brain Tumors from MRI
Segmentation of bone region (from CAT) and the tumor (from NMR) using two fuzzy classifiers.
Category: Computer Assisted Diagnosis
Technologies: 3D Texture Rendering guided by 2D Texture, 3D Registration, Fuzzy Classification, C++.
A tumor begins when normal cells change and grow uncontrollably, forming a mass. A tumor can be benign (noncancerous) or malignant (cancerous). A tumor in the brain can be especially challenging to treat because the person’s thought processes and movements can be affected. A tumor may cause significant symptoms if it grows and presses on the brain or spinal cord.
In this project we have combined two medical imaging modalities. The bone area, shown in green (see figure), was obtained from computed tomography images. The tumor, shown in pink, was obtained from nuclear magnetic resonance imaging. Both images supplied as two-dimensional slices covering the same space. These images have been reconstructed in 3D and a registration process was performed to put in correspondence the anatomical structures provided by both imaging modalities. Two fuzzy classifiers shown in the lower part of the image were used to obtain the bone region (from CAT) and the tumor (from NMR).
Simulated surgical procedure using prior CT-MRI registration...