manisha.aggarwal@jhu.edu Pre-clinical and postmortem diffusion MRI Manisha Aggarwal, Ph.D. Department of Radiology Johns Hopkins University School of Medicine
Diffusion and tissue microstructure? signal 1 1 S = S 0 e b AAA 180 o pulse PGSE 0.8 Diffusion Time ( ) 0.6 slow diffusion 0.4 fast diffusion 0.2 0 0 0 0.5 1 1.5 b-factor 2 connectomethebook.com
Overview Technical considerations for diffusion at high-fields Diffusion signal and brain tissue microstructure? Going beyond pulsed gradients: Oscillating gradient diffusion MRI
Overview Technical considerations for diffusion at high-fields Diffusion signal and brain tissue microstructure? Going beyond pulsed gradients: Oscillating gradient diffusion MRI
Diffusion at high-field: Technical challenges Preclinical MR magnetic field strengths (~4.7-21 T) Stronger field inhomogeneities Shortened tissue T2, T2*, and longer T1 - clinically viable TEs result in SNR loss - requirement of shorter diffusion times (Δ) Eddy current distortions Susceptibility artifacts - limit single-shot EPI applications for faster acquisition
Diffusion at high-fields: Acquisition Acquisition spin echo, fast spin echo (FSE), segmented EPI, GRASE (scan time, resolution, T2-decay, distortion, noise tolerance) Diffusion encoding SE preparation OR Double-refocused SE 90 180 signal 90 180 180 signal RF G TE G RF G TE G δ Δ δ -G -G Reduced eddy currents
Diffusion at high-field: Acquisition Twin navigator echoes for phase error correction 9.4T 11.7T Eddy currents Physiological motion Cai et al. 2011 before after Mori, van Zijl, MRM 1999 Aggarwal et al., MRM 2010
Outline Technical considerations for diffusion at high-fields Diffusion signal and brain tissue microstructure? high angular resolution methods in the brain diffusion beyond white matter effect of spatial resolution for preclinical imaging Going beyond pulsed gradients: Oscillating field gradient diffusion MRI
Diffusion spectrum imaging: resolving crossing fibers in the brainstem Macaque brain 4.7 T 515 points in q-space DSI DTI Wedeen et al., NIMG 2008
Single shell HARDI: crossing fibers using fodfs in the human brainstem fodf TDI FA 30 single-shell directions (11.7T) fodfs using spherical deconvolution (model based) Aggarwal et al., NIMG 2013
Understanding in vivo dmri contrasts *Single-shell with 30 diffusion-sampling orientations 3T 11.7T single-shot 2D EPI 12-shot 3D EPI Shorter T2-1) higher field strength, 2) fixed tissue
Understanding in vivo dmri contrasts b = 1000 s/mm 2 b = 4000 s/mm 2 In vivo ICBM-DTI Ex vivo Fixed brain tissue has lower ADC need to use higher b-values! 2.2 mm resolution 3T 0.25 mm resolution 11.7T
Diffusion contrast and tissue microstructure? Myelin DEC 0.25 x 0.25 x 0.25 mm 3 voxels CST ML 2mm SCP DEC Myelin Nolte s Human brain atlas Mosby 2008
Outline Technical considerations for diffusion at high-fields Diffusion signal and brain tissue microstructure? high angular resolution methods in the brain diffusion beyond white matter effect of spatial resolution for preclinical imaging Going beyond pulsed gradients: Oscillating field gradient diffusion MRI
Diffusion beyond white matter Cortical microstructure in the human brain
HARDI of the human visual cortex Imaging crossing fibers in the Stria of Gennari 242 µm resolution, 60 directions Leuze et al. Cereb Cortex 2014
Intra-cortical layers in the visual cortex Laminar structure using HARDI in the primary visual area V1 30 directions, 92 µm resolution DEC Silver staining Combined high angular and spatial resolution Aggarwal et al., NIMG 2015
Intra-cortical layers in the visual cortex Layered profiles of anisotropy across cortical gray matter Area 17 Area 18 Combined high angular and spatial resolution
Layer-specific fodfs in cortical gray matter T2w DEC Layer III SoG Layer IVc
Can we differentiate cortical areas using HARDI? M1 cortex (BA 4) V1 cortex (BA 17) * * WM WM TDI silver-stain TDI silver-stain single-shell HARDI (30 dir) probabilistic CSD Track density imaging Aggarwal et al. NIMG 2015 Refer to: Tournier et al. NIMG 2007 Calamante et al. NIMG 2010
Outline Technical considerations for diffusion at high-fields Diffusion signal and brain tissue microstructure? high angular resolution methods in the brain diffusion beyond white matter effect of spatial resolution for preclinical imaging Going beyond pulsed gradients: Oscillating field gradient diffusion MRI
Effect of spatial resolution? isotropic voxel size 125 µm 80 µm 55 µm Mouse brain diffusion MRI, ex vivo at 11.7T spatial resolution
Example: tractography in the Reeler mouse cortex In vivo HARDI, 156 x 156 µm 2 resolution, probabilistic tractography Harsan et al., PNAS 2013
Mouse embryonic cortex: towards micro-imaging E15 E15 PSF CP 1 mm IZ z y VZ E18 Phase modulation VZ IZ CP 52-µm resolution, 11.7T ky kz 3D-GRASE
Diffusion and microstructure: understanding diffusion in the developing cortex? Diffusion micro-imaging Fluorescence microscopy E17 E18 Track-density images Radial glia + Microtubules Aggarwal et al., Cereb Cortex 2015
Outline Technical considerations for diffusion at high-fields Diffusion signal and brain tissue microstructure? Going beyond pulsed gradients: oscillating field gradient diffusion MRI
What about diffusion time? PGSE G 180 o pulse G X = Diffusion Time ( ) 2DD Diffusion in the brain is restricted. Can we get additional information by varying the diffusion time?
Diffusion in biological tissues is restricted Temporal diffusion spectrum D (f) Free diffusion Restricted diffusion 0 f (Hz) Measured ADC intrinsic diffusion coefficient of free water
Oscillating gradients for diffusion encoding PGSE (Δ = 10 ms) OGSE f = 50 Hz 0 50 t(ms) OGSE f = 100 Hz 0 50 t(ms) OGSE f = 150 Hz 0 50 t(ms) S= S 0 exp(- F(f)D(f)F(f) df) - 0 50 t(ms) Parsons et al. MRM 2006
Sensitive to varying spatial scales in tissue? Predicted behavior of ADC with increasing frequency Gore et al., NMR Biomed 2010
What about heterogeneous brain tissue microstructure Paxinos mouse brain atlas Baizer, Front Hum Neursci. 2014
OGSE diffusion MRI of the mouse hippocampus T2-weighted ADC Nissl Py GrDG PGSE 50 Hz 100 Hz 150 Hz GrDG 0 1 µm 2 /ms Paxinos Atlas, 2003 Layer-specific contrasts with increasing frequency Aggarwal et al., MRM 2012
OGSE contrasts in the mouse cerebellum 1.0E-3 CBGr T2-w ADC PGSE 50 Hz ADC (mm 2 /s) 8.0E-4 6.0E-4 4.0E-4 CBML 2.0E-4 0 50 100 150 f (Hz) Nissl CBml 100 Hz 150 Hz CBGr CBml CBGr 0 1 µm 2 /ms SMI31 DAPI
Frequency-dependent contrasts in the mouse brain T2w Δ f ADC Nissl 4 CBGr GrDG Δ f ADC 2 Py Pir scx mcx CBml 0 3 µm 2 0
OGSE diffusion MRI in brain tumor Colvin et al., Cancer Res 2008
OGSE diffusion MRI in hippocampal injury - sensitivity of OGSE dmri contrasts to neurodegeneration Aggarwal et al., MRM 2014
Diffusion MRI with circularly-polarized oscillating gradients Fixed monkey cerebellum at 4.7T Lundell et al., MRM 2014
OGSE dmri in the in vivo human brain? corpus callosum Apodized cosine π Trapezoid cosine π 18 Hz 44 Hz 63 Hz Van et al. MRM 2014
OGSE contrast in layers of the human cerebellum Δ f ADC 1.1 0.9 CbGr Cbml WM ADC * * 0.7 0.5 0.3 H&E 0.1 3.5 3 2.5 2 0 67 100 150 200 Hz CbGr Δ f ADC 1.5 1 0.5 Cbml WM 0 4 µm 2 0 Aggarwal et al., ISMRM 2014
Take home messages Diffusion can be sensitized to varying aspects of tissue microstructure Choose acquisition and modeling strategy to best suit the biological question to be answered - DTI vs. HARDI, angular vs. spatial resolution, deterministic vs. probabilistic tracking In addition to b-value and orientation, diffusion time can also play a role dmri using oscillating gradients can reveal additional information about spatial scales in tissue microstructure Questions? manisha.aggarwal@jhu.edu