Project 3

FA1 and FA2
Estimation of FA1 and FA2 of two crossing fiber components at different SNR and separation angles (30, 40, 50 and 80 degrees). (a), (b) and (c) represent different combination of ground truth FA1, FA2 and volume fraction. Dashed lines are true FA values. Green arrows indicate the SNR above which the fitting is accurate and stable.

MTTA: Multi-Tensor Tract based Analysis for clinical diffusion MRI

The diffusion magnetic resonance imaging (MRI) in clinical research indicates the diffusion imaging of b value at 1000-1500s/mm2 and 25 to 60 diffusion orientations with imaging time less than 10 minutes. It is well known that almost half of brain voxels of the human brain contain crossing fibers which make current FA values of single tensor significantly underestimated and cause biased conclusions in clinical applications.

Previous investigations of crossing fibers have been mainly related to correctly revealing the fiber orientations in the voxels of crossing fibers for improvement of tractography. Only recently, several other metrics have been proposed to be an alternate scalar measure at these voxels. In addition, technologies of tract analyses including ours have become important due to great clinical significance of the tracts. However, most of existing metrics for crossing-fiber problem were not designed for tract analysis in that in a single voxel there are no delineation of multi-FA values which are assigned to crossed multi-tracts.

In addition, most of these studies for improving tractography and metrics were performed at the cost of high b, high angular resolution and long scanning time which are not suitable for clinical studies. Due to extensive applications of the FA in neurological and psychiatric studies such as phenotype characterization, drug testing and therapy monitoring, biased conclusions from the underestimated FA in tract analysis could have significantly negative impacts on these studies.

Hence it is critical to develop methods which can correct the underestimated FA in these clinical studies and are designed for tract analysis. With input of diffusion weighted images, MTTA identifies crossing fiber regions first, fully characterizes multiple tensors (a step further than just multiple primary eigenvectors for tractography) at the voxels of crossing fibers and detaches the crossed fibers for tract analysis.

Return to Top