Susanne S. Rauh1, Joep Suskens2, Jithsa R. Monte3, Frank Smithuis3, Oliver J. Gurney-Champion3, Johannes L. Tol2, Mario Maas3, Aart J. Nederveen3, Gustav J. Strijkers1, and Melissa T. Hooijmans3
1Department of Biomedical Engineering and Physics, Amsterdam Movement Sciences, Amsterdam UMC, location AMC, Amsterdam, Netherlands, 2Department of Orthopedic Surgery, Amsterdam UMC, location AMC, Amsterdam, Netherlands, 3Department of Radiology and Nuclear Medicine, Amsterdam UMC, location AMC, Amsterdam, Netherlands
Synopsis
Keywords: Muscle, Diffusion Tensor Imaging
Diffusion tensor imaging (DTI) with correction for
intravoxel incoherent motion (IVIM) is a potential biomarker to assess
hamstring injury recovery and predict return-to-play time. However, the long
acquisition times hinder the use in clinical practice. By accelerating in the
b-value space and using multiband acceleration the scan time can be reduced to
clinically acceptable levels. In this work we showed that those methods
preserve the sensitivity to hamstring injuries and that high-b DTI in
combination with multiband factor 2 can reduce the scan time to 3:40min.
Introduction
Hamstring injuries are common in elite athletes and show
recurrence rates of up to 30%1. Current qualitative imaging
techniques such as T2-weighted MRI fail to predict a return-to-play time (RTP)2. Diffusion tensor imaging (DTI)
is sensitive to microstructural changes in skeletal muscle3,4 and might potentially be able
to assess injury recovery and predict RTP in hamstring injuries. Intra-voxel
incoherent motion causes additional signal decay, corrupting the DTI signal.
Correction for IVIM effects increases the repeatability of DTI5 and yields additional
information about perfusion. However, IVIM-corrected DTI requires long scan
times which limits its clinical applicability.
In this work we aim to accelerate IVIM-corrected DTI of
acute hamstrings injuries to a clinically acceptable scan time while
maintaining its sensitivity. For this purpose, we undersample IVIM-DTI scans in
the b-value space and combine this with multiband (MB) acceleration. Methods
58 patients with an acute hamstring injury (2f/56m, mean age
27.2±8.5
years) underwent MRI examination (3T Ingenia, Philips, Best, The Netherlands)
within 7 days after the injury (baseline). 42 of those patients also received a
follow-up MRI scan within 7 days after their RTP.
The MRI protocol included an anatomical Dixon scan
(1.5x1.5x2.5mm³, 80 slices, 6 echoes) and a SE-EPI IVIM-DTI scan (3x3x5mm³, 40
slices, TE/TR=55ms/5914ms, SPAIR and gradient-reversal fat suppression,
SENSE=1.5) with 10 b-values (range 0-600s/mm², 8x b=0s/mm²) and 56 diffusion
directions. For 7 patients,
the IVIM-DTI scan at baseline and RTP was repeated with MB factor 2 (TE/TR=60ms/3365ms).
The scan times were 11:08min for the IVIM-DTI scan and 6:20min for the MB
IVIM-DTI scan.
The DTI data was denoised and registered to the Dixon images
using QMRITools
6. Injury center slices were
determined by a radiologist and segmentations were drawn manually covering 7
slices in the DTI scan over the center slice in the injured muscle and a
closely matching region in the contralateral healthy muscle in ITK-snap
(www.itksnap.org). The diffusion tensor with IVIM correction was calculated
using nonlinear least-squares fitting in Matlab (R2021a, The MathWorks, Natick,
CA) with two different methods:
- IVIM-corrected
DTI:
This recently proposed method7 consists of four steps:
A. IVIM fit to the mean diffusion-weighted data
B. Voxel-wise IVIM fit with fixed pseudo-diffusion coefficient D* obtained from A.
C. Subtraction of the IVIM-component from the full signal.
D. DTI fit using all b-values to the remaining signal. - High-b
DTI:
We undersampled the data by discarding all b-values<200s/mm², resulting in
b=200,400,600s/mm² and 32 diffusion directions to estimate the diffusion
tensor. Assuming that the IVIM signal has fully decayed at b=200s/mm², the
tensor is inherently IVIM corrected. The scan time is 6:27min without MB and 3:40min
with MB.
Additionally,
we estimated the IVIM perfusion fraction f with a two-step IVIM-DTI fit, fixing
the diffusion tensor calculated with high-b DTI, and thereafter using all
b-values for the IVIM fit.
The
outcome parameters for both methods were mean diffusivity (MD), fractional
anisotropy (FA), diffusion tensor eigenvalues λ
1, λ
2, λ
3,
and IVIM-perfusion fraction (f). The Δ(injured-healthy) values per parameter
and method were calculated as a measure of sensitivity to injury at baseline
and RTP. A Wilcoxon signed rank test was used to assess differences in
sensitivity between fitting methods and MB acceleration for each of the DTI parameters.
Repeated measures Anova was used to assess if differences in injury state and
between MB and non-MB parameters were significant (p≤0.05).
Results
Outcome parameters for fully sampled scans and accelerated
scans using the various approaches were not significantly different (Figure 1).
With high-b DTI the sensitivity to hamstring injuries could be preserved for
all parameters both, with and without MB acceleration, compared to the
IVIM-corrected DTI fit using all b-values (p>0.05, Figure 2).
At baseline, significant changes between injured and healthy
muscle were found in λ1, λ2, λ3 and MD with all
methods, but not for FA and f. Interestingly, significant differences were also
found at RTP; however, Figure
3 shows
that the injured values converged towards healthy values at RTP. For all
parameters and both injury states, no differences between MB and no-MB outcome
parameters were found. Discussion
In this work we demonstrated that sensitivity to hamstring
injuries is preserved with our accelerated high-b DTI method and high-b DTI combined
with MB undersampling. High-b DTI with MB factor 2 enables a scan time of less
than 5 minutes, which makes this protocol suitable for clinical practice.
FA and f did not show differences between injury states,
indicating that those parameters are not sensitive to hamstring injuries. For the
IVIM parameter estimation, we averaged over the diffusion directions and did
not correct for T2 effects which could negatively impact the estimation.
Therefore, incorporating a T2 correction8 or taking into account the
anisotropy of the capillary network in skeletal muscle9 might be useful in the future.
In the current setting, sampling of small b-values for IVIM parameter
estimation seems redundant and high-b DTI in combination with MB factor 2 can
be regarded the method of choice in hamstring injuries. Conclusion
High-b DTI combined with MB factor
2 reduces the scan time to a clinically acceptable level and preserves the
sensitivity to hamstring injuries. Acknowledgements
This study was partially funded by the ‘National Basketball Association (NBA) &
General Electric Healthcare (GEHC) Sports Medicine Collaboration, USA’. References
1. Orchard J, Best TM. The management of
muscle strain injuries: An early return versus the risk of recurrence. In: Clinical Journal of Sport Medicine. Vol 12. ;
2002:3-5. doi:10.1097/00042752-200201000-00004
2. Reurink
G, Brilman EG, de Vos RJ, et al. Magnetic Resonance Imaging in Acute Hamstring Injury:
Can We Provide a Return to Play Prognosis? Sport
Med. 2014;45(1):133-146. doi:10.1007/s40279-014-0243-1
3. Froeling
M, Oudeman J, Strijkers GJ, et al. Muscle changes detected with diffusion-tensor imaging
after long-distance running. Radiology. 2015;274(2):548-562.
doi:10.1148/radiol.14140702
4. Hooijmans MT, Monte JRC, Froeling M, et
al. Quantitative MRI Reveals Microstructural Changes in the Upper Leg Muscles
After Running a Marathon. J Magn Reson Imaging. 2020.
doi:10.1002/jmri.27106
5. Monte JR, Hooijmans MT, Froeling M, et
al. The repeatability of bilateral diffusion tensor imaging (DTI) in the upper
leg muscles of healthy adults. Eur Radiol. 2019. doi:10.1007/s00330-019-06403-5
6. Froeling M. QMRTools: a Mathematica
toolbox for quantitative MRI analysis. J Open Source Softw.
2019;4(38):1204. doi:10.21105/joss.01204
7. Schlaffke L, Rehmann R, Rohm M, et al.
Multi‐center evaluation of stability and reproducibility of quantitative MRI
measures in healthy calf muscles. NMR Biomed. July 2019.
doi:10.1002/nbm.4119
8. Jerome NP, D’Arcy JA, Feiweier T, et
al. Extended T2-IVIM model for correction of TE dependence of pseudo-diffusion
volume fraction in clinical diffusion-weighted magnetic resonance imaging. Phys
Med Biol. 2016;61(24):N667-N680. doi:10.1088/1361-6560/61/24/N667
9. Karampinos DC, King KF, Sutton BP,
Georgiadis JG. Intravoxel partially coherent motion technique: Characterization
of the anisotropy of skeletal muscle microvasculature. J Magn Reson Imaging.
2010;31(4):942-953. doi:10.1002/jmri.22100