A Preliminary Study of DTI Fingerprinting on Stroke Analysis.

Heather T. Ma,Chenfei Ye,Jun Wu,Pengfei Yang,Xuhui Chen,Zhengyi Yang,Jingbo Ma
DOI: https://doi.org/10.1109/embc.2014.6944100
2014-01-01
Abstract:DTI (Diffusion Tensor Imaging) is a well-known MRI (Magnetic Resonance Imaging) technique which provides useful structural information about human brain. However, the quantitative measurement to physiological variation of subtypes of ischemic stroke is not available. An automatically quantitative method for DTI analysis will enhance the DTI application in clinics. In this study, we proposed a DTI Fingerprinting technology to quantitatively analyze white matter tissue, which was applied in stroke classification. The TBSS (Tract Based Spatial Statistics) method was employed to generate mask automatically. To evaluate the clustering performance of the automatic method, lesion ROI (Region of Interest) is manually drawn on the DWI images as a reference. The results from the DTI Fingerprinting were compared with those obtained from the reference ROIs. It indicates that the DTI Fingerprinting could identify different states of ischemic stroke and has promising potential to provide a more comprehensive measure of the DTI data. Further development should be carried out to improve DTI Fingerprinting technology in clinics.
What problem does this paper attempt to address?