Validation of Diffusion Tensor MRI with Structure Tensor Synchrotron Imaging
Irvin Teh1, Darryl McClymont1, Marie-Christine Zdora2,3, Valentina Davidoiu4, Hannah J Whittington1, Christoph Rau2, Irene Zanette2, and Jürgen E Schneider1

1Division of Cardiovascular Medicine, Radcliffe Department of Medicine, University of Oxford, Oxford, United Kingdom, 2Diamond Light Source, Didcot, United Kingdom, 3Department of Physics and Astronomy, University College London, London, United Kingdom, 4Department of Imaging Sciences and Biomedical Engineering, King's College London, London, United Kingdom

Synopsis

Diffusion tensor imaging (DTI) is widely used to assess tissue microstructure, but is limited in resolution and cannot resolve multiple fibre populations within a voxel. Existing methods for validating DTI are limited in either resolution or coverage. 2D histological methods are additionally destructive and prone to tissue distortion. In contrast, synchrotron imaging strikes an excellent balance between resolution and coverage. Here, we demonstrate for the first time, the prospect of validating DTI with structure tensor analysis of synchrotron imaging data. Tensors reconstructed with DTI and structure tensor synchrotron imaging were consistent across the left ventricular wall of the heart.

Purpose

Diffusion tensor imaging (DTI) is increasingly used to investigate the 3D tissue architecture in the heart. It enables estimation of the principal orientations of cells and sheets based on the principal eigenvectors ν1, ν2 and ν3. However, the process of diffusion in biological tissue is complicated by the presence of cell membranes, organelles and metabolites. Independent validation is critical to improving interpretation of DTI data, and broadening its application. Previous validation studies made use of ink prints1 and microscopic histology.2,3 However, these 2D approaches are prone to tissue distortion during sample preparation and are destructive. Reconstructing the data in 3D in the presence of distortions remains an unresolved challenge. Methods for extracting tissue structural information in 3D include anatomical MRI4 and CT with X-ray laboratory sources5 which are limited in resolution, scanning electron microscopy6 and confocal microscopy7 which are limited in field-of-view (FOV), and synchrotron imaging (SI) which strikes an excellent balance between resolution and coverage.8,9 Here, we investigate for the first time, the prospect of validating DTI with structure tensor (ST) analysis of SI data.

Methods

One heart was excised from a healthy female Sprague-Dawley rat, fixed in isosmotic Karnovsky’s fixative, doped with 2mM Gd and embedded in a tube of agarose gel for MRI and SI. Non-selective 3D fast spin echo DTI data were acquired on a 9.4 T preclinical MRI scanner (Agilent, CA, USA) with a transmit-receive birdcage coil (Rapid Biomedical, Rimpar, Germany) of inner diameter = 20 mm. TR / TE1 = 250 / 9.5 ms, echo spacing = 5.1 ms, echo train length = 8, FOV = 21.6 x 14.4 x 14.4 mm, resolution (isotropic) = 100 μm, number of non-DW images = 4, number of DW directions = 30, b = 1,000 s/mm2. SI was subsequently performed at beamline I13-2 (imaging branch) of the Diamond Light Source (Didcot, UK) using a polychromatic beam with X-ray energies in the interval 20 - 30 keV. Data were acquired with (A) 1.25X objective, FOVreconstructed = 14.0 x 14.0 x 9.6 mm, pixel size = 3.6 μm, and (B) 4X objective, FOVreconstructed = 4.5 x 4.5 x 3.0 mm, pixel size = 1.1 μm. Diffusion tensors were fitted to the DTI data, and diffusion primary eigenvectors (ν1,DT) were mapped. STs were calculated based on the signal gradient intensity in the SI absorption data, and transformed such that the primary eigenvectors (ν1,ST) were oriented along the dominant cell long axes. Helix angles were calculated as the elevation angle of ν1,DT and ν1,ST perpendicular to the axial plane. All data were analysed with custom code in Matlab R2013A (Mathworks, Natick, USA).

Results

Figure 1 depicts (A) diffusion primary eigenvector (ν1,DT) maps and (B) SI absorption data in a short-axis mid-myocardial slab. The data exhibit substantial 3D coverage, isotropic image resolution and excellent anatomical correspondence at the macroscopic scale free from distortion. Closer inspection of ν1,DT in a 3D region-of-interest (ROI), as in Figure 2A, reveals the known transmural variation in helix angle, with the color scheme encoding for ν1,DT orientation. Figure 2B illustrates the superior resolution of SI, and its ability to resolve cellular structures. Diffusion and structure tensors in 2D ROIs across the septal wall (Figure 3) are in good agreement, and recapitulate the transmural variation in myocyte orientation, where helix angles progress from negative to positive values from the subepicardium to the subendocardium.

Discussion

DTI is a valuable method that enables inference of 3D tissue structure in the whole intact organ. Previous efforts to validate DTI using other high resolution imaging methods are routinely limited by FOV or resolution, or are susceptible to tissue distortion during sample preparation. SI overcomes these constraints. In conjunction with ST analysis, it has both the resolution for assessing cellular microstructure, and the coverage for measuring macroscopic fibre and sheet architecture in the whole heart. Synchrotron radiation is known for its high brilliance, and provides orders of magnitude higher flux than X-ray laboratory sources, enhancing the achievable resolution. ST-SI can readily yield tensors at a higher resolution than DTI. This would help disentangle complex structural information that would have been averaged over the larger voxels used in DTI. ST-SI is uniquely placed to serve to validate and aid development of DTI and advanced diffusion MRI methods, and also to inform mechanical modelling of biological tissue. In further work, we will obtain SI data in additional hearts with whole organ coverage, optimize the acquisition parameters and improve tissue contrast via phase retrieval.

Acknowledgements

This work was supported by the EPSRC, UK (EP/J013250/1), BBSRC, UK (BB/I012117/1) and the British Heart Foundation Centre for Research Excellence, UK (FS/11/50/29038; PG/13/33/30210). The authors acknowledge a Wellcome Trust Core Award (090532/Z/09/Z). We thank Diamond Light Source for access to beamline I13-2 (MT13044-1) that contributed to the results presented here.

References

1. Tseng WY, Wedeen VJ, Reese TG, Smith RN, Halpern EF. Diffusion tensor MRI of myocardial fibers and sheets: correspondence with visible cut-face texture. J Magn Reson Imaging 2003;17(1):31-42.

2. Holmes AA, Scollan DF, Winslow RL. Direct histological validation of diffusion tensor MRI in formaldehyde-fixed myocardium. Magn Reson Med 2000;44(1):157-161.

3. Kung GL, Nguyen TC, Itoh A, Skare S, Ingels NB, Jr., Miller DC, Ennis DB. The presence of two local myocardial sheet populations confirmed by diffusion tensor MRI and histological validation. J Magn Reson Imag 2011;34(5):1080-1091.

4. Bernus O, Radjenovic A, Trew ML, LeGrice IJ, Sands GB, Magee DR, Smaill BH, Gilbert SH. Comparison of diffusion tensor imaging by cardiovascular magnetic resonance and gadolinium enhanced 3D image intensity approaches to investigation of structural anisotropy in explanted rat hearts. J Cardiovasc Magn Reson 2015;17(1):31.

5. Ni H, Castro SJ, Stephenson RS, Jarvis JC, Lowe T, Hart G, Boyett MR, Zhang H. Extracting Myofibre Orientation from Micro-CT Images: An Optimisation Study. In: Proceedings of the 40th Annual Conference on Computing in Cardiology, Zaragoza, Spain, 2013.

6. LeGrice IJ, Smaill BH, Chai LZ, Edgar SG, Gavin JB, Hunter PJ. Laminar structure of the heart: ventricular myocyte arrangement and connective tissue architecture in the dog. Am J Physiol 1995;269(2 Pt 2):H571-582.

7. Young AA, LeGrice IJ, Young MA, Smaill BH. Extended confocal microscopy of myocardial laminae and collagen network. J Microsc 1998;192(Pt 2):139-150.

8. Varray F, Wang L, Fanton L, Zhu YM, Magnin IE. High Resolution Extraction of Local Human Cardiac Fibre Orientations. In: Proceedings of the 7th International Conference on Functional Imaging and Modeling of the Heart, London, United Kingdom, 2013.

9. Balicevic V, Loncaric S, Cárdenes R, Gonzalez-Tendero A, Paun B, Crispi F, Butakoff C, Bijnens B. Assessment of Myofiber Orientation in High Resolution Phase-Contrast CT Images. In: Proceedings of the 8th International Conference on Functional Imaging and Modeling of the Heart, Maastricht, The Netherlands, 2015.

Figures

Figure 1. (A) Orthogonal views of DTI primary eigenvector (ν1,DT) map in an ex vivo rat heart. A 3D region-of-interest (ROI) is magnified in Figure 2 (yellow box). (B) Matching 3D synchrotron absorption data in a 14 x 14 x 9.6 mm section with pixel size = 3.6 μm.

Figure 2. (A) Orthogonal views of ν1,DT in a 3 x 3 x 3 mm section in the septal wall. Three 2D ROIs are magnified in Figure 3 (blue, yellow and red boxes). (B) Matching 3D synchrotron absorption data with pixel size = 1.1 μm show fine structural detail.

Figure 3. Diffusion and structure tensors (DT & ST) within 0.5 x 0.5 x 0.1 mm mid-axial sections in the epicardial, mid-myocardial and endocardial septal wall. The tensors are displayed at 0.1 mm resolution, in axial (ax) and coronal (cor) views and color coded by helix angle (HA).



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
0922