<html>
  <head>
    <meta content="text/html; charset=ISO-8859-1"
      http-equiv="Content-Type">
  </head>
  <body bgcolor="#FFFFFF" text="#000000">
    <div class="moz-cite-prefix">Hello,<br>
      <br>
      the artefact added with eddy_correct is more pronounce for high
      bvalue (3000) with 1000 or 1500 it should go better<br>
      For the bvec correction it is a tiny effect so it does not matter
      that much, (that's why it is not in there priority list), if you
      run eddy_correct you can use the extra script fdt_rotate_bvecs <br>
      <br>
      The eddy procedure is much better : this is mainly because in
      addition to motion it also correct the distorsion due to eddy
      current. (and also EPI distorsion with topup). The main limitation
      is that it require a specific acquisition (either the direction
      have to be over the all sphere or you need each direction with
      direct and oposite phase direction), but It worth to change your
      dwi acquisition.<br>
      <br>
      I hope it helps <br>
      <br>
      Romain<br>
      &nbsp;<br>
      Le 18/11/2013 23:39, Jeurissen Ben a &eacute;crit&nbsp;:<br>
    </div>
    <blockquote
      cite="mid:384E4D541D819449ABBBD507434317D190F0B7A6@xmail34.ad.ua.ac.be"
      type="cite">
      <meta http-equiv="Content-Type" content="text/html;
        charset=ISO-8859-1">
      <style type="text/css" id="owaParaStyle"></style>
      <style type="text/css"></style>
      <div style="direction: ltr;font-family: Tahoma;color:
        #000000;font-size: 10pt;">
        <div>
          <div style="font-family:Tahoma; font-size:13px">
            <div style="font-family:Tahoma; font-size:13px">
              <div style="font-family:Tahoma; font-size:13px">
                <div style="widows:2; text-transform:none;
                  text-indent:0px; letter-spacing:normal;
                  font-style:normal; font-variant:normal;
                  font-size:13px; line-height:normal;
                  font-family:Tahoma; white-space:normal; orphans:2;
                  word-spacing:0px">
                  Hi all,</div>
                <div style="widows:2; text-transform:none;
                  text-indent:0px; letter-spacing:normal;
                  font-style:normal; font-variant:normal;
                  font-size:13px; line-height:normal;
                  font-family:Tahoma; white-space:normal; orphans:2;
                  word-spacing:0px">
                  <br>
                </div>
                <div style="widows:2; text-transform:none;
                  text-indent:0px; letter-spacing:normal;
                  font-style:normal; font-variant:normal;
                  font-size:13px; line-height:normal;
                  font-family:Tahoma; white-space:normal; orphans:2;
                  word-spacing:0px">
                  FSL's "eddy_correct" doesn't do b-matrix rotation.
                  They do offer a separate script to reorient the bvecs
                  file after running eddy_correct, though.</div>
                <div style="widows:2; text-transform:none;
                  text-indent:0px; letter-spacing:normal;
                  font-style:normal; font-variant:normal;
                  font-size:13px; line-height:normal;
                  font-family:Tahoma; white-space:normal; orphans:2;
                  word-spacing:0px">
                  <br>
                </div>
                <div style="widows:2; text-transform:none;
                  text-indent:0px; letter-spacing:normal;
                  font-style:normal; font-variant:normal;
                  font-size:13px; line-height:normal;
                  font-family:Tahoma; white-space:normal; orphans:2;
                  word-spacing:0px">
                  FSL5's "eddy" doesn't do b-matrix rotation either and
                  there is no script available to reorient the bvecs
                  file. On their mailinglist the developers stated
                  developing such a tool is currently not high on their
                  priority list.</div>
                <div style="widows:2; text-transform:none;
                  text-indent:0px; letter-spacing:normal;
                  font-style:normal; font-variant:normal;
                  font-size:13px; line-height:normal;
                  font-family:Tahoma; white-space:normal; orphans:2;
                  word-spacing:0px">
                  <br>
                </div>
                <div style="widows:2; text-transform:none;
                  text-indent:0px; letter-spacing:normal;
                  font-style:normal; font-variant:normal;
                  font-size:13px; line-height:normal;
                  font-family:Tahoma; white-space:normal; orphans:2;
                  word-spacing:0px">
                  Hope this clarifies the state of b-matrix rotation in
                  FSL.</div>
                <div style="widows:2; text-transform:none;
                  text-indent:0px; letter-spacing:normal;
                  font-style:normal; font-variant:normal;
                  font-size:13px; line-height:normal;
                  font-family:Tahoma; white-space:normal; orphans:2;
                  word-spacing:0px">
                  <br>
                </div>
                <div style="widows:2; text-transform:none;
                  text-indent:0px; letter-spacing:normal;
                  font-style:normal; font-variant:normal;
                  font-size:13px; line-height:normal;
                  font-family:Tahoma; white-space:normal; orphans:2;
                  word-spacing:0px">
                  Cheers,</div>
                <div style="widows:2; text-transform:none;
                  text-indent:0px; letter-spacing:normal;
                  font-style:normal; font-variant:normal;
                  font-size:13px; line-height:normal;
                  font-family:Tahoma; white-space:normal; orphans:2;
                  word-spacing:0px">
                  Ben</div>
                <div style="widows:2; text-transform:none;
                  text-indent:0px; letter-spacing:normal;
                  font-style:normal; font-variant:normal;
                  font-size:13px; line-height:normal;
                  font-family:Tahoma; white-space:normal; orphans:2;
                  word-spacing:0px">
                  <br>
                </div>
                <div style="widows:2; text-transform:none;
                  text-indent:0px; letter-spacing:normal;
                  font-style:normal; font-variant:normal;
                  font-size:13px; line-height:normal;
                  font-family:Tahoma; white-space:normal; orphans:2;
                  word-spacing:0px">
                  <div>-<br>
                    <div>
                      <div>
                        <div>
                          <div>
                            <font size="1">Ben Jeurissen, Ph.D.</font></div>
                          <div>
                            <div><font size="1">Post-doctoral researcher</font></div>
                            <div><font size="1">Vision Lab, Dept. of
                                Physics</font></div>
                            <div><font size="1">University of Antwerp</font></div>
                            <div><font size="1">Universiteitsplein 1,
                                N.1.18</font></div>
                            <div><font size="1">B-2610 Wilrijk, Belgium</font></div>
                            <div><font size="1">Phone: +32 3 265 24 77</font></div>
                            <div><font size="1">Email:
                                <a class="moz-txt-link-abbreviated" href="mailto:ben.jeurissen@ua.ac.be">ben.jeurissen@ua.ac.be</a></font></div>
                            <div><font size="1">Url:&nbsp;<a
                                  moz-do-not-send="true"
                                  title="http://visielab.ua.ac.be/people/ben-jeurissen
                                  Ctrl+Click to follow link"
                                  href="http://visielab.ua.ac.be/people/ben-jeurissen"
                                  target="_blank">http://visielab.ua.ac.be/people/ben-jeurissen</a></font></div>
                          </div>
                        </div>
                      </div>
                    </div>
                  </div>
                </div>
              </div>
            </div>
          </div>
        </div>
        <div style="font-family: Times New Roman; color: #000000;
          font-size: 16px">
          <hr tabindex="-1">
          <div id="divRpF964956" style="direction: ltr;"><font
              color="#000000" face="Tahoma" size="2"><b>From:</b>
              <a class="moz-txt-link-abbreviated" href="mailto:mrtrix-discussion-bounces@public.nitrc.org">mrtrix-discussion-bounces@public.nitrc.org</a>
              [<a class="moz-txt-link-abbreviated" href="mailto:mrtrix-discussion-bounces@public.nitrc.org">mrtrix-discussion-bounces@public.nitrc.org</a>] on behalf of
              Ivan Alvarez [<a class="moz-txt-link-abbreviated" href="mailto:ivan.alvarez.11@ucl.ac.uk">ivan.alvarez.11@ucl.ac.uk</a>]<br>
              <b>Sent:</b> 18 November 2013 13:31<br>
              <b>To:</b> Donald Tournier<br>
              <b>Cc:</b> mrtrix mailinglist<br>
              <b>Subject:</b> Re: [Mrtrix-discussion] Motion correction
              best practise<br>
            </font><br>
          </div>
          <div>
            <div class="moz-cite-prefix">Hi Donald,<br>
              <br>
              Thank you for all your comments. <br>
              <br>
              I was not aware FSL did b-matrix update, will definitely
              look into this. <br>
              <br>
              It's interesting that you mention we may be loosing more
              than we are gaining with motion correction, particularly
              with the introduction of artefacts and poor registration
              at high b-values. Is there a rule of thumb for when are
              such procedures useful/detrimental? For example, in fMRI
              movement &gt;5mm within a single run (i.e. 5-10 minutes
              continuous scanning) is considered worrisome but
              salvageable and &gt;10mm is often too much to be rescued
              by registration. Is there something like that for dMRI?<br>
              <br>
              PS The Gaussian process code is now in FSL under the name
              EDDY ( <a moz-do-not-send="true"
                href="http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/EDDY"
                target="_blank">
                http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/EDDY</a>). I
              haven't personally tried it, but would love to hear from
              other users!<br>
              <pre class="moz-signature" cols="72">Kind regards,
Ivan Alvarez

PhD Candidate
Imaging and Biophysics Unit
UCL Institute of Child Health
30 Guilford Street, London, WC1N 1EH</pre>
              On 18/11/13 12:15, Donald Tournier wrote:<br>
            </div>
            <blockquote type="cite">
              <div dir="ltr">Hi Ivan,
                <div><br>
                </div>
                <div>I'm probably not the best person to ask about
                  this... There's a few other people on this list whose
                  opinion on the matter is much better informed than
                  mine, I cordially invite them all to chip in. ;)</div>
                <div><br>
                </div>
                <div>I've no experience with ExploreDTI, but you can ask
                  Alex Leemans directly through the ExploreDTI mailing
                  list. That said, I'm fairly confident it does the
                  b-matrix update...</div>
                <div><br>
                </div>
                <div>As to FSL, as far as I know it also does
                  volume-wise affine registration, and it is possible to
                  do the b-matrix update, although I'm not familiar with
                  the procedure needed to do this.&nbsp;</div>
                <div><br>
                </div>
                <div>I'm not familiar with any package that does
                  slice-by-slice registration, but my gut feeling on the
                  matter is that there's probably not a great deal of
                  point to doing this as slice-wise mis-registration is
                  generally accompanied by through-plane motion, which
                  will cause signal corruption due to spin history
                  effects. For this reason, I'd consider the entire
                  volume affected to be corrupt in this case: even if
                  there is no obvious signal drop, the chances are there
                  will be some corruption. That said, it might be worth
                  doing if your downstream processing pipeline has a way
                  of handling outliers, etc.&nbsp;</div>
                <div><br>
                </div>
                <div>I'd also like to highlight that these approaches
                  are still far from perfect. I've already raised the
                  issue on this list, but based on my limited exposure
                  to FSL's eddy_correct (which seems to be what most
                  people use), I think it often creates more problems
                  than it solves. I hasten to add that this problem may
                  also apply to other approaches, but so far I've only
                  been exposed to data processed using eddy_correct.
                  I've come across many cases where the coregistration
                  introduces artefacts, even when the original data
                  wasn't particularly affected in the first place. These
                  artefacts typically consist of the DW images being
                  stretched along one or more axes, probably because the
                  algorithm tries to match the parenchyma bit of the DW
                  volumes to the parenchyma+CSF parts of b=0 volume.
                  This is particularly pronounced with high b-value
                  data, but I've also seen it happen in run-of-the-mill
                  b=1000 data too (as recently as a couple of weeks ago,
                  in fact). All this to say, if you use these tools,
                  please don't treat them as a black box, do check that
                  they're working as expected.&nbsp;</div>
                <div><br>
                </div>
                <div>On the upside, Jesper Anderson recently proposed a
                  new approach based on Gaussian processes, which I
                  think has now made it into FSL5. If any other users
                  have tried using it, feel free to comment on this...</div>
                <div><br>
                </div>
                <div>Cheers,</div>
                <div>Donald.</div>
                <div><br>
                </div>
              </div>
              <div class="gmail_extra"><br>
                <br>
                <div class="gmail_quote">On 18 November 2013 03:06, Ivan
                  Alvarez <span dir="ltr">
                    &lt;<a moz-do-not-send="true"
                      href="mailto:ivan.alvarez.11@ucl.ac.uk"
                      target="_blank">ivan.alvarez.11@ucl.ac.uk</a>&gt;</span>
                  wrote:<br>
                  <blockquote class="gmail_quote" style="margin:0 0 0
                    .8ex; border-left:1px #ccc solid; padding-left:1ex">
                    <div bgcolor="#FFFFFF">Hi Donald,<br>
                      <br>
                      I wanted to bring up motion correction again,
                      particularly what is recommended for and against
                      in MRtrix. I am aware the issue has been raised in
                      the mailing list before, but it might be useful to
                      have an idea of what is generally a good or bad
                      idea.<br>
                      <br>
                      So far, I have seen people doing
                      motion/eddy-current correction in either
                      ExploreDTI or FSL. The documentation for both is
                      these is somewhat scant and I am trying to piece
                      together what do they exactly do. This is my naive
                      reading so far, please feel free to correct me:<br>
                      <br>
                      ExploreDTI<br>
                      * Affine registration<br>
                      * Whole volume at a time<br>
                      * Updates B-matrix<br>
                      <br>
                      FSL<br>
                      * Affine registration<br>
                      * Slice-by-slice<br>
                      * Does<i> not</i> update B-matrix<br>
                      <br>
                      &gt;From what I understand in the discussion, the
                      slice-by-slice registration is preferable to avoid
                      smearing artefacts across the whole volume while
                      updating the B-matrix can generally improve
                      results (<a moz-do-not-send="true"
                        href="http://www.ncbi.nlm.nih.gov/pubmed/19319973"
                        target="_blank">http://www.ncbi.nlm.nih.gov/pubmed/19319973</a>).

                      Is this roughly correct? If so, are there any
                      other considerations specific to MRtrix?<span
                        class="HOEnZb"><font color="#888888"><br>
                          <pre cols="72">-- 
Kind regards,
Ivan Alvarez

PhD Candidate
Imaging and Biophysics Unit
UCL Institute of Child Health
30 Guilford Street, London, WC1N 1EH </pre>
                        </font></span></div>
                    <br>
                    _______________________________________________<br>
                    Mrtrix-discussion mailing list<br>
                    <a moz-do-not-send="true"
                      href="mailto:Mrtrix-discussion@www.nitrc.org"
                      target="_blank">Mrtrix-discussion@www.nitrc.org</a><br>
                    <a moz-do-not-send="true"
                      href="http://www.nitrc.org/mailman/listinfo/mrtrix-discussion"
                      target="_blank">http://www.nitrc.org/mailman/listinfo/mrtrix-discussion</a><br>
                    <br>
                  </blockquote>
                </div>
                <br>
                <br clear="all">
                <div><br>
                </div>
                -- <br>
                <div dir="ltr"><b><font color="#990000">Dr J-Donald
                      Tournier (PhD)</font></b><br>
                  <div><font color="#990000"><br>
                    </font></div>
                  <i><font color="#990000">Senior Lecturer,&nbsp;</font></i><i><font
                      color="#990000">Biomedical Engineering</font></i>
                  <div><i><font color="#990000">Division of Imaging
                        Sciences &amp; Biomedical Engineering<br>
                        King's College London</font></i>
                    <div><i><font color="#990000"><br>
                        </font></i></div>
                    <div><i><font color="#990000"><b
                            style="font-family:Calibri,sans-serif;
                            font-size:15px"><span style="font-size:10pt">A:</span></b><span
                            style="font-family:Calibri,sans-serif;
                            font-size:10pt">&nbsp;Department of Perinatal
                            Imaging &amp; Health, 1<sup>st</sup>&nbsp;Floor
                            South Wing, St Thomas' Hospital, London. SE1
                            7EH</span><br>
                        </font></i></div>
                    <div><i><font color="#990000"><b>T:</b> +44 (0)20
                          7188 7118 ext 53613</font></i></div>
                  </div>
                  <div><i><font color="#990000"><b>W:</b>&nbsp;<a
                          moz-do-not-send="true"
href="http://www.kcl.ac.uk/medicine/research/divisions/imaging/departments/biomedengineering"
                          target="_blank">http://www.kcl.ac.uk/medicine/research/divisions/imaging/departments/biomedengineering</a></font></i><br>
                  </div>
                </div>
              </div>
            </blockquote>
            <br>
          </div>
        </div>
      </div>
      <br>
      <fieldset class="mimeAttachmentHeader"></fieldset>
      <br>
      <pre wrap="">_______________________________________________
Mrtrix-discussion mailing list
<a class="moz-txt-link-abbreviated" href="mailto:Mrtrix-discussion@www.nitrc.org">Mrtrix-discussion@www.nitrc.org</a>
<a class="moz-txt-link-freetext" href="http://www.nitrc.org/mailman/listinfo/mrtrix-discussion">http://www.nitrc.org/mailman/listinfo/mrtrix-discussion</a>
</pre>
    </blockquote>
    <br>
  </body>
</html>