Multiband (aka simultaneous multislice) EPI validation in progress!

I am pleased to see a couple of presentations at next week's ISMRM conference in Salt Lake City dealing with some of the important validation steps that should be performed before multiband (MB) EPI (or simultaneous multislice (SMS) EPI if you prefer) is adopted for routine use by the neuroimaging community:

Characterization of Artifactual Correlation in Highly-Accelerated Simultaneous Multi-Slice (SMS) fMRI Acquisitions

Abstract #0410, ISMRM Annual Meeting, 2013.

Kawin Setsompop, Jonathan R. Polimeni, Himanshu Bhat, and Lawrence L. Wald

Simultaneous Multi-Slice (SMS) acquisition with blipped-CAIPI scheme has enabled dramatic reduction in imaging time for fMRI acquisitions, enabling high-resolution whole-brain acquisitions with short repetition times. The characterization of SMS acquisition performance is crucial to wide adoption of the technique. In this work, we examine an important source of artifact: spurious thermal noise correlation between aliased imaging voxels. This artifactual correlation can create undesirable bias in fMRI resting-state functional connectivity analysis. Here we provide a simple method for characterizing this artifactual correlation, which should aid in guiding the selection of appropriate slice- and inplane-acceleration factors for SMS acquisitions during protocol design.

(I also found this link to the full abstract.)

An Assessment of Motion Artefacts in Multi Band EPI for High Spatial and Temporal Resolution Resting State fMRI

Abstract #3275, ISMRM Annual Meeting, 2013.

Michael E. Kelly, Eugene P. Duff, Janine D. Bijsterbosch, Natalie L. Voets, Nicola Filippini, Steen Moeller, Junqian Xu, Essa S. Yacoub, Edward J. Auerbach, Kamil Ugurbil, Stephen M. Smith, and Karla L. Miller

Multiband (MB) EPI is a recent MRI technique that offers increased temporal and/or spatial resolution as well as increased temporal SNR due to increased temporal degrees-of-freedom (DoF). However, MB-EPI may exhibit increased motion sensitivity due to the combination of short TR with parallel imaging. In this study, the performance of MB-EPI with different acceleration factors was compared to that of standard EPI, with respect to subject motion. Although MB-EPI with 4 and 8 times acceleration exhibited some motion sensitivity, retrospective clean-up of the data using independent component analysis was successful at removing artefacts. By increasing temporal DoF, accelerated MB-EPI supports higher spatial resolution, with no loss in statistical significance compared to standard EPI. MB-EPI is therefore an important new technique capable of providing high resolution, temporally rich FMRI datasets for more interpretable mapping of the brain's functional networks.

The natural question to ask next occurs at the interface of these two topics: what about head motion-driven artifactual correlations between simultaneously excited slices? I am also curious to see how retrospective motion correction, e.g. affine registration algorithms, performs with MB-EPI that contains appreciable motion contamination. Is the "pre-processing" pipeline that we use for single-shot EPI appropriate for MB-EPI?

In-plane parallel imaging such as GRAPPA and SENSE were adopted for EPI-based fMRI experiments prematurely in my view, i.e. before full validations had been conducted. (Mea culpa. I was one of those beguiled by GRAPPA when I first saw it.) The failure modes - like motion sensitivity - hadn't been fully explored before a lot of us began employing the methods for their purported benefits. It would be nice if the failure modes of MB-EPI get a thorough workout before the neuroimaging community adopts it en masse. 

That said, I am still very excited that MB-EPI may offer the most significant performance boost for fMRI acquisition for more than a decade (since the introduction of scanners capable of EPI readout on all three gradient axes). But I continue to seek validation before recommending widespread adoption of MB-EPI (or any other method) and I look forward to seeing more reports such as these in the literature and online, prior to people using them in experiments to solve the brain.

Resting state fMRI confounds

(Thanks to Dave J. Hayes for tweeting the publication of these papers.)

Two new papers provide comprehensive reviews of some of the confounds to the acquisition, processing and interpretation of resting state fMRI data. In the paper, "Resting-state fMRI confounds and cleanup," Murphy, Birn and Bandettini consider in some detail many of the noise sources in rs-fMRI, especially those having a physiologic origin.

In "Overview of potential procedural and participant-related confounds for neuroimaging of the resting state," Duncan and Northoff review the effects that other circumstantial factors, such as the scanner's acoustic noise, subject instructions, subjects' emotional state, and caffeine might have on rs-fMRI studies. Without due consideration, some or all of these factors may inadvertently become experimental variables; the implications for inter-individual differences are considerable. (I've reviewed some of the issues concerning what we can permit subjects to do before and during rs-fMRI in this post.)

While we're on the subject of confounds in rs-fMRI - especially those with a motion component - another confound that motion introduces is a sensitivity to the receive field heterogeneity of the head coil. This problem gets worse the more channels the coil has, because the coil elements get smaller as the number of channels goes up. For an introduction to the issue see this arXiv paper; there will also be simulations of the effect for a 32-channel coil at the ISMRM conference in a couple of weeks' time. (See e-poster, abstract #3352.) The result is that spurious correlations and anti-correlations can result, necessitating some sort of clever sorting or de-noising scheme to distinguish them from "true" brain correlations. I mention it here because there is a common misconception in the field that applying a retrospective motion correction step fixes all motion-related artifacts. It doesn't. Nor does including all of the motion parameters as regressors in a model. Motion has some insidious ways in which it can modulate the MRI signal level, and it is high time that we, as a field, reconsider very carefully what we are doing for motion correction, and why.

Finally, I'll note in passing that slice timing correction may not be a good idea for rs-fMRI. It's been known since the correction was first proposed that it should interact a with a motion correction step. (The two corrections should be applied simultaneously, as one 4D space-time correction rather than a separate 3D space then time correction, or vice versa.) I don't have data to share just yet, but if anyone is wondering whether they should include STC in their rs-fMRI analysis, as they would do for event-related fMRI, then my advice is to skip it until someone can prove to you that it has no unintended consequences. (Demonstration of unintended consequences to follow eventually....)


References:

Resting state fMRI confounds and cleanup. K Murphy, RM Birn and PA Bandettini, NeuroImage Epub.
DOI: 10.1016/j.neuroimage.2013.04.001

Overview of potential procedural and participant-related confounds for neuroimaging of the resting state. NW Duncan and G Northoff, J. Psychiatry Neurosci. 2013, 38(2), 84-96.
PMID: 22964258
DOI: 10.1503/jpn.120059

Impressively rapid follow-ups to a published fMRI study

Alternative post title: Why blogs can be seriously useful in research.


Last week there was quite a lot of attention to an article published in PNAS by Aharoni et al. In their study they claimed that fMRI could be useful in predicting the likelihood of rearrest in a group of convicts up for parole:

"Identification of factors that predict recurrent antisocial behavior is integral to the social sciences, criminal justice procedures, and the effective treatment of high-risk individuals. Here we show that error-related brain activity elicited during performance of an inhibitory task prospectively predicted subsequent rearrest among adult offenders within 4 y of release (N = 96). The odds that an offender with relatively low anterior cingulate activity would be rearrested were approximately double that of an offender with high activity in this region, holding constant other observed risk factors. These results suggest a potential neurocognitive biomarker for persistent antisocial behavior."

The senior author, Kent Kiehl, was interviewed on National Public Radio on Friday morning. I heard it on my way into work. An NPR interview would suggest the media attention was widespread, although I haven't looked at this aspect specifically.

What I did notice, however, was that The Neurocritic came out with two quick posts (here and here) wherein he brought up a couple of interesting limitations of the study and even ran his own re-analysis of the data, the PNAS authors having been kind enough to make their data available publicly.

This afternoon, Russ Poldrack has followed up with his own analysis and interpretation of the study's data. I'll be honest, all the stats leaves me flat-footed. But I am very seriously impressed by the way the blogosphere, combined with shared data, has been able to poke and prod the original study's conclusions.

Why am I so enthused? Because the mainstream media (still) has the power to dominate the narrative in the public sphere, and it is especially important that specific criticisms can be leveled within the same news cycle, while the public might still be paying attention to the story. So, while I think it's highly unlikely that NPR will interview the senior author of the next study that finds there is no predictive use of fMRI for recidivism - we seem to have a serious positive results bias in science - maybe there's a slim chance that NPR will interview Russ about his follow-up analysis, just to balance the record. And if not, at least those in the field have the benefit of the post-publication peer review that blogs can offer.

Quick update for Siemens users

Apologies for the lengthy absence. Many irons in the fire, etc. So until I can provide a more considered post I give you these three random tidbits:

1. Syngo MR version D13 for Verio and Skyra

There is an EPI sequence in VD13 that has a real time update of the on-resonance frequency, i.e. one that is computed and applied TR by TR, to combat drift caused by gradient heating. There are apparently versions for fMRI and diffusion-weighted imaging. I don't have any detailed information but if you are working on a Verio or a Skyra it might be time to talk to your physicist and/or local Siemens rep.

2. Phase encode direction for axial and axial-oblique EPI

Siemens uses A-P phase encoding by default whereas GE uses P-A by default. Essentially, for axial (and axial oblique) EPI the A-P direction compresses the frontal lobe but stretches occipital lobe whereas P-A stretches frontal lobe and compresses occipital. Pick your poison. (See Note 1.) Test each one out by setting the Phase enc. dir. parameter on the Routine tab. To set P-A from A-P (default) first click the three dots (...) to the right of the parameter field and open the dialog box, then enter 180 <return> instead of 0. You will probably find that the parameter change doesn't "stick" for appended scans, so saving a modified protocol in the Exam Explorer is a way to ensure the default (A-P) doesn't get reinstated without you noticing. More details to come in the next version of my user training/FAQ document.

3. Another way to force a re-shim

In my last user training/FAQ document (and here) I gave a simple way to force the scanner to re-shim at any point, e.g. when you know or strongly suspect the subject may have moved, or between lengthy blocks as a way to maintain high quality data in spite of slow subject motion and scanner drifts. But there is another way to do it and from some basic tests it looks to be superior. Here's a shaky video of the procedure conducted on a Trio running Syngo MR B17 (see Note 2):

(The essential procedure is the same for later software versions, but the layout of the 3D Shim window is slightly different.)

A checklist for fMRI acquisition methods reporting in the literature

This post updates the draft checklist that was presented back in October. Thanks to all who provided feedback. The updated checklist, denoted version 1.1, incorporates a lot of the suggestions made previously. The main difference is the reduction from three to two categories. The logic is that we should be encouraging reporting of "All of Essential plus any of Supplemental" parameters in the methods section of any fMRI publication.

(Click to enlarge.)

Explanatory notes, consolidated from the post on the draft list, and abbreviations appear below.

Inadequate fat suppression for diffusion imaging

Diffusion imaging is often included as a component of functional neuroimaging protocols these days. While fMRI examines functional changes on the timescale of seconds to minutes, diffusion imaging is able to detect changes over weeks to years. Furthermore, there may be complimentary information from the white matter connectivity obtainable from diffusion imaging – that is, from tractography - and the functional connectivity of gray matter regions that can be derived from resting state or task-based fMRI experiments.

I was recently made aware of some artifacts on diffusion-weighted EPI scans acquired on a colleagues’ scanner. When I was able to replicate the issue on my own scanner, and even make the problem worse, it was time to do a serious investigation. The origin of the problem was finally confirmed after exhaustive checks involving the assistance of several engineers and scientists from Siemens. The conclusion isn't exactly a major surprise: fat suppression for diffusion-weighted imaging of brain is often insufficient. And it seems that although the need for good fat suppression is well known amongst physics types, it’s not common knowledge in the neuroscience community. What’s more, the definition of “sufficient” may vary from experiment to experiment and it may well be that numerous centers are unaware that they may have a problem.

Let’s start out by assessing a bad example of the problem. The diffusion-weighted images you’re about to see were acquired from a typical volunteer on a Siemens TIM/Trio using a 32-channel receive-only head coil, with b=3000 s/mm² (see Note 1), 2 mm isotropic voxels, and GRAPPA with twofold (R=2) acceleration. These are three successive axial slices:

(Click to enlarge.)

The blue arrows mark hypointense artifacts whereas the orange arrow picks out a hyperintense artifact. Even my knowledge of neuroanatomy is sufficient to recognize that these crescents are not brain structures. They are actually fat signals, shifted up in the image plane from the scalp tissue at the back of the head. (If you look carefully you may be able to trace the entire outline of the scalp, including fat from around the eye sockets, all displaced anterior by a fixed amount.) I’ll discuss the mechanism later on, but at this point I’ll note that the two principal concerns are the b value (of 3000 s/mm²) and the use of a 32-channel array coil. GRAPPA isn’t a prime suspect for once!

Now, part of the problem is that the intensity of the artifacts – but not their location - changes as the direction of the diffusion-weighting gradients changes. In the following video you see the diffusion-weighted images as the diffusion gradient orientation is changed through thirty-two directions (see Note 2):

The signal from white matter fibers changes as the diffusion gradient direction changes. That’s what you want to happen. But the displaced fat artifacts also change intensity with diffusion gradient direction, meaning that the artifact is erroneously encoded as regions of anisotropic diffusion. Thus, when one computes the final diffusion model, the brain regions contaminated by fat artifacts end up looking like white matter tracts. In the next figure the data shown above was fit to a simple tensor model, from which a color-coded anisotropy map can be obtained:

The white arrow picks out the false “tract” corresponding to the artifact signal crescent we saw on the raw diffusion-weighted images. I suppose it’s remotely possible that this is the iTract, a new fasciculus that has evolved to connect the subject’s ear to his smart phone, but my money is on the fat artifact explanation.

Clearly, in the above image there is no easy way to distinguish the artifact from real white matter tracts by eye, except by using your prior anatomical knowledge. And it's likely to confuse tractographic methods, too, because it has very similar geometric properties to those that tractographic methods attempt to trace. So let's take a look at the origin of the problem and then we can get into what you want: solutions. 

Review: Differentiating BOLD from non-BOLD signals in fMRI time series using multi-echo EPI

Disclaimer: I'm afraid I haven't done a very good job reviewing the entirety of this paper because the stats/processing part was pretty much opaque to me. I've done my best to glean what I can out of it, and then I've focused as much as I can on the acquisition, since that is one part where I can penetrate the text and offer some useful commentary. Perhaps someone with better knowledge of stats/ICA/processing will review those sections elsewhere.


The last paper I reviewed used a bias field map to attempt to correct for some of the effects of subject motion in time series EPI. A different approach is taken by Prantik Kundu et al. in another recently published study. In their paper, Differentiating BOLD from non-BOLD signals in fMRI time series using multi-echo EPI, Kundu et al. set out to differentiate between signal changes that have a plausible neurally-driven BOLD origin from those that are likely to have been modulated by something other than neuronal activity. In the latter category we have cardiac and respiratory fluctuations and, of course, subject motion.

The method involves sorting BOLD-like from spurious changes using an independent component analysis (ICA) and to then "de-noise" the time series before applying connectivity analysis. For resting state fMRI in particular, the lack of any sort of ground truth and an absence of independent knowledge that one has with task-based fMRI makes disambiguating neurally driven signal changes from artifacts a major problem. Kundu et al. use a relatively simple philosophical approach to the separation:

"We hypothesized that if TE-dependence could be used to differentiate BOLD and non-BOLD signals, non-BOLD signal could be removed to denoise data without conventional noise modeling. To test this hypothesis, whole brain multi-echo data were acquired at 3 TEs and decomposed with Independent Components Analysis (ICA) after spatially concatenating data across space and TE. Components were analyzed for the degree to which their signal changes fit models for R₂* and S₀ change, and summary scores were developed to characterize each component as BOLD-like or not BOLD-like."

And, noting again the caveat that there is an absence of ground truth, the approach seems to work:

"These scores clearly differentiated BOLD-like “functional network” components from non BOLD-like components related to motion, pulsatility, and other nuisance effects. Using non BOLD-like component time courses as noise regressors dramatically improved seed-based correlation mapping by reducing the effects of high and low frequency non-BOLD fluctuations."