Kao Lun Wang, Presenter: Nothing to Disclose

Shiou-Ping Lee, Abstract Co-Author: Nothing to Disclose

Li Liu, Abstract Co-Author: Nothing to Disclose

Wen-Yih Tseng, Abstract Co-Author: Nothing to Disclose

Geometric distortion of echo-planar diffusion weighted images comes from eddy current- and susceptibility-induced artifact. The aim of this work is to utilize the existing SPM2 built-in coregistration functions based on similarity metric methods and implement them in raw DWIs to correct for shift and linear distortion. Our major goal is to find the optimum algorithm in the existing methods to obtain more accurate tractographic results.

A healthy volunteer were scanned on a 1.5T MRI system (Siemens). A transaxial-diffusion SE EPI sequence was used to acquire 55 slices covering whole brain. Isotropic voxel size = 2.2x2.2x2.2 mm3. DWI b value = 1000 s/mm2, TR/TE = 8900/95 ms. Diffusion DTI 13-encoding schemes acquire for a DTI data about 7 minutes.

We corrected DWI data using four different SPM2 built-in coregistration functions, i.e., normalized mutualinformation(NMI), mutual-information(MI) entropy correlation coefficient (ECC)and normalized correlation coefficient(NCC), and compared the reconstructed tractograpahic results between them. We showed that tractographic results reconstructed under NMI coregistration were consistent with known anatomy, supporting a previous report that using NMI spatial transformation model was a robust and efficient method to correct for raw DWI. Tractography of corpus callosum was used to compare among raw DWI, corrected data using ECC, NCC and MI in terms of coherence index (CI). The coherence index was a measure of similarity of an individual fiber tract with respect to the gold standard tractography; here the NMI model was used. We found that CI in raw and NCC were 143.90(182.90) and 673.45(654.90), respectively, which were significantly higher than CI in ECC, 20.32(28.40), and MI, 20.28(27.58).

Our study shows that ECC and MI are as effective as NMI in the correction of DWI distortion. The diffusion tensor and tractographic results reconstructed under ECC or MI coregistration are comparable and closer to those reconstructed under NMI coregistration.

Our major goal is to find the optimum algorithm in the existing methods to obtain more accurate tractographic results to help relative to clinical study.

Wang, K, Lee, S, Liu, L, Tseng, W, Assessment of Existing Coregistration Methods for Distortion Correction and Their Effects on Diffusion Tensor Reconstruction and Tractography.  Radiological Society of North America 2007 Scientific Assembly and Annual Meeting, November 25 - November 30, 2007 ,Chicago IL. http://archive.rsna.org/2007/5001243.html