In this section
@article{Giannakidis:2016:21/7765,
author = {Giannakidis, A and Melkus, G and Yang, G and Gullberg, GT},
doi = {21/7765},
journal = {Physics in Medicine and Biology},
title = {On the Averaging of Cardiac Diffusion Tensor MRI Data: The Effect of Distance Function Selection},
url = {http://dx.doi.org/10.1088/0031-9155/61/21/7765},
volume = {61},
year = {2016}
}
TY - JOUR
AB - Diffusion tensor magnetic resonance imaging (DT-MRI) allows a unique insightinto the microstructure of highly-directional tissues. The selection of the mostproper distance function for the space of diffusion tensors is crucial in enhancingthe clinical application of this imaging modality. Both linear and nonlinearmetrics have been proposed in the literature over the years. The debate on themost appropriate DT-MRI distance function is still ongoing. In this paper, we presenteda framework to compare the Euclidean, affine-invariant Riemannian andlog-Euclidean metrics using actual high-resolution DT-MRI rat heart data. We employedtemporal averaging at the diffusion tensor level of three consecutive andidentically-acquired DT-MRI datasets from each of five rat hearts as a means torectify the background noise-induced loss of myocyte directional regularity. Thisprocedure is applied here for the first time in the context of tensor distance functionselection. When compared with previous studies that used a different concrete applicationto juxtapose the various DT-MRI distance functions, this work is uniquein that it combined the following: (i) Metrics were judged by quantitative –ratherthan qualitative– criteria, (ii) the comparison tools were non-biased, (iii) a longitudinalcomparison operation was used on a same-voxel basis. The statisticalanalyses of the comparison showed that the three DT-MRI distance functions tendto provide equivalent results. Hence, we came to the conclusion that the tensormanifold for cardiac DT-MRI studies is a curved space of almost zero curvature. The signal to noise ratio dependence of the operations was investigated throughsimulations. Finally, the “swelling effect” occurrence following Euclidean averagingwas found to be too unimportant to be worth consideration.
AU - Giannakidis,A
AU - Melkus,G
AU - Yang,G
AU - Gullberg,GT
DO - 21/7765
PY - 2016///
SN - 1361-6560
TI - On the Averaging of Cardiac Diffusion Tensor MRI Data: The Effect of Distance Function Selection
T2 - Physics in Medicine and Biology
UR - http://dx.doi.org/10.1088/0031-9155/61/21/7765
VL - 61
ER -
For enquiries about the MRI Physics Collective, please contact:
Mary Finnegan
Senior MR Physicist at the Imperial College Healthcare NHS Trust
Pete Lally
Assistant Professor in Magnetic Resonance (MR) Physics at Imperial College
Jan Sedlacik
MR Physicist at the Robert Steiner MR Unit, Hammersmith Hospital Campus