Diffusion MRI signals and artifacts

1. Diffusion and Brownian motion

Diffusion is the microscopic movement of a substance without bulk motion.

The self-diffusion of water molecules follows the Brownian motion.

source: https://en.wikipedia.org/wiki/Brownian_motion


1. What is the difference between perfusion and diffusion?

2. Diffusion coefficient, diffusion time, and diffusion distance

Diffusivity measures the rate of diffusion

Fick's first law of diffusion: the substance transfer J through an area is proportional to the gradient of concentration c (i.e. δc/δx): 

where D is diffusivity, also termed diffusion coefficient, has a unit of mm²/s. The higher the diffusivity, the faster (or high rate) they diffuse.

Einstein [1] derived the root mean squared displacement of a self-diffusion as , where D is diffusivity, and d is also termed the diffusion distance. One should note that diffusivity is more like a velocity measurement (distance over time) not a quantity measurement (quantity over time). High diffusivity implies faster diffusion, not necessarily a large quantity of diffusion.


Diffusivity does not quantify the amount of diffusion
Assumes (1) no structural restriction (2) homogeneity in the diffusion environment. 


The diffusion coefficient, a.k.a diffusivity, is a velocity measurement quantifying how fast substance diffuses.
Diffusion distance: the root-mean-square displacement of free self-diffusion.


1. How to quantify diffusion?
2. What does diffusivity measure?
3. What is the diffusion distance at a diffusion time of 32 ms? (the self-diffusion coefficient of water at the body temperature is 3x10³ mm²/s = 3 micron²/ms)

3. Diffusion NMR 

Diffusion MR measures diffusivity by signal attenuation

The effect of diffusion on the MR signal was mentioned by Hahn in the 1950s [2]. A more popular approach, the Stejskal-Tanner sequence, also known as the pulse-gradient spin-echo (PGSE) sequence, was developed in the 1960s to measure the diffusion coefficient [4](Fig. 1A). The modified version of this sequence is still used today in most of clinical scanners. The diffusion signals (S) and diffusivity (D) have the following relation.

The final signal for a spin-echo diffusion sequence after adding T1 and T2 weighting is thus:

Dr. Le Bihan used a "b factor" to simplify the diffusion signal pattern (Le Bihan et al., 1986):

b-value is a collective product of parameters controlling the diffusion sensitization (i.e., sensitivity to diffusion).

δ: diffusion-encoding gradient duration
Δ: the diffusion time
g: diffusion-encoding gradient
γ: gyromagnetic ratio
S0: diffusion signal without diffusion weighting (the b0 signal)

The diffusion coefficient can thus be measured by acquiring diffusion signals with difference gradient strength |g|. '

Different sequence designs are available to acquire the diffusion-weighted images.

Fig.1 (A) the Stejskal-Tanner sequence (B) Diffusion-encoded stimulated echo sequence. The diffusion time (Δ) equals TE/2 plus the mixing time (TM) and is thus much longer than TE. (C) Stimulated echo sequence with bipolar diffusion-encoding gradients. Bipolar gradients lengthen the TE significantly and become limiting. (D) The Stejskal-Tanner sequence with bipolar diffusion-encoding gradients.
source: Sosnovik, D. E., Wang, R., Dai, G., Reese, T. G., & Wedeen, V. J. (2009). Diffusion MR tractography of the heart. Journal of Cardiovascular Magnetic Resonance, 11(1), 1-15. [link]

The diffusion-encoding gradient can have different forms:
(1) magnitude: monopolar, bipolar, oscillating (e.g. Oscillating Gradient Spin Echo (OGSE) diffusion sequence)
(2) direction: single fixed direction, trajectories (q-trajectory imaging)
(3) duration: long encoding duration, pulse gradient duration


The self-diffusion coefficient can be measured using an NMR experiment.


Assumes free diffusion.


1. Diffusion sequence uses gradients to encode/sensitize diffusion.
2. Diffusion is presented by an exponential signal attenuation.
3. The diffusion signal attenuation is determined by the diffusion coefficient, diffusion time, and the strength/duration of the diffusion sensitization gradient.


1. What are the relation between diffusion time, diffusion encoding gradient, and diffusion signals?
2. What is the relation between diffusion signals and TE, TR? 
3. Is the diffusion signal mostly T1- or T2-weighted? Why? (T1 = time of the recovery of the longitudinal magnetization after excitation by an RF pulse. T2 = the spin dephasing time in the transverse plane after RF excitation)
4. Why does a longer diffusion time require a longer TE in the Stejskal-Tanner PGSE sequence?
5. Scan A has TE=100 ms whereas scan 2 has TE=130 ms. What is the SNR difference in diffusion MRI at the white matter? (white matter T2 = 69 ms, T1 = 1080 ms, Mxy = Mxy0 exp(-TE/T2) )
6. A study uses capillary tubes to create a diffusion phantom, but the diameter of capillary tubes is larger than axons. How would you adjust the diffusion time and diffusion gradient strength?
7. In ex-vivo tissues, the diffusivity drops substantially. What parameters have to be adjusted in comparison with in-vivo tissue.

4. Diffusion MRI

Diffusion weighted images are spatial distributions of water weighted by diffusion attenuation.

The combination of a diffusion sequence and k-space imaging makes diffusion MRI possible. The PGSE sequence developed in the 1960s allowed for measuring the diffusivity of free water diffusion using an NMR experiment, and it was yet to be combined with MRI (Lauterbur, 1973) to spatially map the distribution of diffusion coefficient. This idea gave birth to diffusion-weighted imaging at the mid-1980s [5], and Michael Moseley introduced its important role in the diagnosis of ischemic stroke [6].


Diffusion MRI allows for spatial mapping of the diffusion signals. 


Due to the additional diffusion gradients, diffusion MRI often has prominent eddy current artifact, susceptibility artifact, and fat saturation artifact (note that these artifacts also exist in other types of MRI sequence).

5. Distortion and artifact

Diffusion MRI is known to have large eddy current distortion and phase distortion.

Susceptibility distortion and artifact

It is noteworthy that susceptibility not only causes distortion due to phase shift but also signal dropout. 

The distortion can be corrected using FSL TOPUP by using two acquisitions acquired at opposite phase encoding directions. However, FSL TOPUP cannot handle signal dropout due to susceptibility. Thus the effective correction strategy still needs to combine sequence setting and post-processing to minimize susceptibility distortion and artifact.

Diffusion gradient encoding increases echo time and result in greater distortion. A longer diffusion time also increases echo time. To reduce the artifact: (1) Increase bandwidth: Using a stronger gradient coil to reduce encoding duration with the expense of SNR. (2) Reduce TE: use fast imaging methods to reduce TE, or use a stimulated echo sequence (But SNR is reduced by half in stimulated echo). (3) use a read-out segmented EPI.

source: https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/topup/TopupUsersGuide

Eddy current distortion

Eddy current induced by gradient switch can affect the signal readout, leading to distorted images [7]. The solutions are (1) correcting the partial distortion by a linear transformation, (2) using a bipolar pulse for diffusion encoding (Fig. 1D).

source: http://www.diffusion-imaging.com/2012/03/dti-preprocessing-distortion-correction.html

Fat saturation artifact 

Fat saturation artifact becomes prominent as b-value increases. Fat signals can be greatly reduced using fat suppression. However, diffusion MRI relies on signal attenuation (high b-value, low signal) to quantify the diffusion coefficient. In high b-value images, signals from water are gone while the fat signals remain. Consequently, it causes errors in diffusion analysis. The solution is to use better fat saturation settings.

The figure shows the fat saturation artifact in b0 (b) and FA map (c).
source: Yeh, P. H., Oakes, T. R., & Riedy, G. (2012). Diffusion tensor imaging and its application to traumatic brain injury: basic principles and recent advances.

Quality control
1. Check the consistency of b-table, TE, and spatial resolutions.
2. Compute the correlation between DWIs acquired by similar diffusion encoding gradients
3. Checking slice-wise signal dropout
4. Check the orientation of the b-table using fiber coherence index.

Correction using FSL Eddy

The most popular tool for correcting eddy current and phase distortion artifact is FSL eddy. see https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/eddy

FSL eddy includes the following corrections:

If no TOPUP result:
1. Register all DWIs to b0 to correct for eddy current artifact and subject movement
2. Register slice to the volume within each DWI (optional) to correct slice-specify eddy current artifact
3. Correct signal dropout in slices

If TOPUP result supplied:
1. Restrict registration when registering all DWIs to b0
2. Restrict registration when registering slices to volumes (optional)
3. Correct susceptibility artifact
3. Correct signal dropout in slices

Eddy works for single-based diffusion data and does NOT work for non-shell-based data (e.g. DSI grid data). 

1. Diffusion MRI usually suffers from the susceptibility artifact and eddy current artifact.
2. The eddy current artifact can be corrected using a software tool (provided that the SNR is good enough) or a set of specially designed diffusion gradients.

1. What is the relation between diffusion gradient strength, diffusion time, and eddy current artifact, susceptibility artifact, fat saturation artifact in diffusion MRI?
2. How to correct for eddy current distortion? Is there any limitation?
3. How to correct for phase distortion artifact? Is there any limitation?

6. Exercise

1. Load an example diffusion MRI data set at https://pitt.box.com/v/course-artifact . Create an SRC in DSI Studio and open it in STEP2 Reconstruction to see the raw DWI.
2. Point out where eddy current distortion, phase distortion, fat saturation artifacts are.
3. Compare the b0 with one DWI. Which regions brighten up? Why?
4. Use DSI Studio to run a quality control assessment on DWI data.

7. Reference

[1] Einstein, Albert. Investigations on the Theory of the Brownian Movement. Courier Corporation, 1956. (pdf)
[2] E L Hahn (1950). Spin echoes. Physical Review, 80:580–594. 
[3] Carr, H. Y., & Purcell, E. M. (1954). Effects of diffusion on free precession in nuclear magnetic resonance experiments. Physical review, 94(3), 630. [link]
[4] Stejskal, E. O., & Tanner, J. E. (1965). Spin diffusion measurements: spin echoes in the presence of a time‐dependent field gradient. The journal of chemical physics, 42(1), 288-292. [link]
[5] Le Bihan, D., Breton, E., Lallemand, D., Grenier, P., Cabanis, E., & Laval-Jeantet, M. (1986). MR imaging of intravoxel incoherent motions: application to diffusion and perfusion in neurologic disorders. Radiology, 161(2), 401-407. [link]
[6] Moseley, M. E., J. Kucharczyk, J. Mintorovitch, Y. Cohen, J. Kurhanewicz, N. Derugin, H. Asgari, and D. Norman. "Diffusion-weighted MR imaging of acute stroke: correlation with T2-weighted and magnetic susceptibility-enhanced MR imaging in cats." American Journal of Neuroradiology 11, no. 3 (1990): 423-429. (link)
[7] Jezzard, P., Barnett, A. S., & Pierpaoli, C. (1998). Characterization of and correction for eddy current artifacts in echo planar diffusion imaging. Magnetic resonance in medicine, 39(5), 801-812.

8. Reference

*NMR is a physical phenomenon in which atomic nuclei in a magnetic field absorb and re-emit electromagnetic radiation. The resonance frequency depends on the strength of the magnetic field and the magnetic properties of the nuclei.

*Spin echo

source: https://en.wikipedia.org/wiki/Spin_echo