Fusing Structural, Functional and Diffusion Tensor MR Image for Neurological Disorders
Structural Magnetic Resonance Imaging (sMRI) provides high-quality images of soft tissue through the use external magnetic fields and electromagnetic radio-frequency pulses to excite protons abundant in the human body. Diffusion Tensor MRI (dtMRI) is a unique, non-invasive imaging technique capable of measuring the anisotropic diffusion of water molecules in biological tissues. The resulting image reflects both the tissue structure (including fiber orientation) and the architecture at the microscopic level which makes it suitable for observing the development in the human cerebral white matter. dtMRI imaging modality results in a 3D field of 2nd order rank 3 tensors where a 3×3 symmetric positive definite matrix is associated with each voxel. The eigenvalues of the tensor at each voxel give the magnitude of diffusion of water molecules at that voxel and its eigenvectors depict the diffusion direction. Standard image processing and analysis techniques are not useful for dtMRI images due to the different data at each voxel and this is the challenge being faced and new techniques for averaging, smoothing, segmentation, visualization and analysis of such images are being developed. Functional MRI (fMR) produces images highlighting active areas of the brain by measuring quantities proportional to the level of oxygenation in the blood. The different MR sub-modalities have presented several challenging problems which occupied researchers in different disciplines including medicine, computing, mathematics, statistics, and visualization during the past few decades. In this project, the intern proposes to develop mathematical and computational algorithms to extract, align, and fuse complementary neurological disease information from the different modalities allowing for mapping brain wiring (DTMRI) and communication (fMRI) over anatomy (sMRI).
