Skip to main content

Directions vs. averages: an in-vivo comparison for cardiac DTI


The ability to interrogate cardiac microstructure has led to much recent interest in in-vivo cardiac diffusion tensor imaging (cDTI). However, when compared to studies performed in neuro-imaging, very little work has been done to determine the optimal diffusion encoding schemes. Previous work has suggested that accuracy is improved by increasing the number of diffusion encoding directions(Ndirs)1,2, but comparisons in the heart have been limited to fixed animal specimens3. Here we compare parameters derived from cDTI using data acquired in vivo with an increasing Ndirs.


10 healthy subjects were imaged on a Siemens Skyra using the STEAM-EPI cDTI sequence4 in a short-axis slice of the mid left-ventricle with the optimal protocol recently described5 (b-values: 150 and 750 smm-2, 2.8x2.8x8mm3 resolution). This was repeated with Ndirs=6, 10, 12 and 20 (standard Siemens product directions) and 12 averages (Navs) were acquired in each direction. The diffusion tensor and parameter maps including mean diffusivity (MD), helical angle (HA) and fractional anisotropy (FA) were calculated as previously described5, using all averages and all directions together to provide a reference data set. The processing was then repeated for each set of diffusion encoding directions with varying numbers of averages chosen to match the total images used Ntot = Navs x Ndirs, as closely as possible to 24, 36 and 60 and also using Navs=12.


Figure 1 shows example parameter maps (HA, MD and FA) calculated using all directions (Ndir=48) and all averages (Nav=12) together compared to each diffusion encoding scheme processed with Ntot=60. There was no consistently visible difference between the encoding schemes. Figure 2 shows mean MD and FA values for each Ndirs plotted with the Ntot and the average variation (standard deviation) in these parameter maps over the left ventricle. For a given Ntot, Ndirs=10 appears to have the minimum variation of FA across the left ventricle and most commonly has FA closest the reference value and Ndirs=12 appears to be superior when considering MD. However, a comparison of MD and FA values when Ntot=60, showed no statistically significant difference between Ndirs (1-way repeated measures ANOVA; MD: p=0.59; FA: p=0.82).

Figure 1

Maps of helical angle (top row), fractional anisotropy (middle) and mean diffusivity (bottom row) for one example healthy subject. Maps were calculated for one reference data set (left hand column) with all available data and then with combinations of Ndirs and Nav to total Ntot=60 in each case. No consistent changes in parameter map were observed when altering Ndirs.

Figure 2

Mean values of fractional anisotropy (A) and mean diffusivity (B) for all 10 subjects (smaller points indicate standard deviations) in the left ventricle, colour coded by the Ndirs and plotted with the Ntot used in the tensor calculation. The horizontal lines indicate the mean values obtained from the reference data set which used every encoding direction and Nav=12. The lower subplots show the standard deviation in the left ventricle averaged over all 10 subjects to provide some indication of the variability of the parameters within a subject. For a given Ntot, the data acquired with Ndirs=10 directions appears to most commonly have the least variation and the FA closest to the reference value. For MD, Ndirs=12 appears to be optimal.


While simulations in previous work have found increasing Ndirs to result in more accurate results, our results suggest that any resultant changes in MD or FA measured in in-vivo myocardium are small.


This work was performed at The National Institute for Health Research Funded Cardiovascular Biomedical Research Unit at The Royal Brompton Hospital and Imperial College London.


  1. 1.

    Jones : MRM. 2004, DOI:10.1002/mrm.20033

    Google Scholar 

  2. 2.

    Ennis : MIA. 2014, DOI: 10.1016/

    Google Scholar 

  3. 3.

    Mazumder : JCMR. 2014, doi:10.1186/1532-429X-16-S1-P359

    Google Scholar 

  4. 4.

    Nielles-Vallespin : MRM. 2012, DOI:10.1002/mrm.24488

    Google Scholar 

  5. 5.

    Scott : MRM. 2014, DOI:10.1002/mrm.25418

    Google Scholar 

Download references

Author information



Corresponding author

Correspondence to Andrew D Scott.

Rights and permissions

This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. The Creative Commons Public Domain Dedication waiver ( applies to the data made available in this article, unless otherwise stated.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Scott, A.D., Ferreira, P., Nielles-Vallespin, S. et al. Directions vs. averages: an in-vivo comparison for cardiac DTI. J Cardiovasc Magn Reson 17, P25 (2015).

Download citation


  • Fractional Anisotropy
  • Diffusion Tensor Imaging
  • Diffusion Tensor
  • Mean Diffusivity
  • Helical Angle