<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
  </head>
  <body>
    <p>Hi Joel,</p>
    <p>The standard init_type ("mni") does a registration between the
      T1w MNI template and the input scan using a correlation metric.
      This assumes that the tissue intensities of the input scan and the
      MNI template are more or less the same, and it can sometimes fail
      if that's not the case (even if the input is a T1w scan). <br>
    </p>
    <p>Init_type="atlas" does a registration between the input and a
      probabilistic atlas using a mutual information-based metric
      (similar to what SPM12 does). This metric doesn't care about the
      specific tissue intensities of the input, so you could, for
      example, feed in a T2w scan and it should still work.</p>
    <p>The reason I set "mni" as default is that it's faster and seemed
      to work well on the cases I tested it. I'll try to do more testing
      on different T1w scans to see if we should set the "atlas"-option
      as default.</p>
    <p>Best,</p>
    <p>Oula</p>
    <p>  <br>
    </p>
    <div class="moz-cite-prefix">On 8/5/22 18:12, Joel Upston wrote:<br>
    </div>
    <blockquote type="cite"
      cite="mid:194A1B83-E82B-4B57-AE1A-00517C2FF857@unm.edu">
      <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
      <div dir="ltr"> Oula,
        <div><br>
        </div>
        <div>Thank you for the quick response. Yes the template
          alignment seems to be a bit better from the affine from
          headreco, than that of what is given through the charm
          optimization. I tried the noneck which didn’t change it but
          just tried the “atlas” init_type parameter and that worked.
          How exactly is it different? <br>
          <div><br>
            <div dir="ltr">Thank you,
              <div><br>
              </div>
              <div>Joel Upston</div>
            </div>
            <div dir="ltr"><br>
              <blockquote type="cite">On Aug 5, 2022, at 5:18 AM, Oula
                <a class="moz-txt-link-rfc2396E" href="mailto:oulap@drcmr.dk"><oulap@drcmr.dk></a> wrote:<br>
                <br>
              </blockquote>
            </div>
            <blockquote type="cite">
              <div dir="ltr">
                <p class="MsoNormal"
                  style="line-height:12.0pt;background:#FF6666"><b><span
                      style="font-size:10.0pt;color:white">  [EXTERNAL]</span></b></p>
                <div>
                  <p>Hi Joel,</p>
                  <p>Did you check the template alignment after forcing
                    the affine from headreco? You can also try switching
                    the init_type from "mni" to "atlas" in the [samseg]
                    section of the settings file. It also looks like the
                    scans don't have neck so please include the --noneck
                    flag if that's the case.</p>
                  <p>If you've tried all these things already, I'm happy
                    to have a look if you're allowed to share a scan or
                    two from the data set.</p>
                  <p>Best,<br>
                    Oula</p>
                  <p><br>
                  </p>
                  <div class="moz-cite-prefix">On 8/5/22 12:32, Joel
                    Upston wrote:<br>
                  </div>
                  <blockquote type="cite"
                    cite="mid:222E2ED5-F642-4E69-84BB-1131664D4FE8@unm.edu">
                    <div dir="ltr">
                      <div style="font-family: Calibri, Arial,
                        Helvetica, sans-serif; font-size: 12pt; color:
                        rgb(0, 0, 0);" class="elementToProof">
                        Hi there,</div>
                      <div style="font-family: Calibri, Arial,
                        Helvetica, sans-serif; font-size: 12pt; color:
                        rgb(0, 0, 0);" class="elementToProof">
                        <br>
                      </div>
                      <div style="font-family: Calibri, Arial,
                        Helvetica, sans-serif; font-size: 12pt; color:
                        rgb(0, 0, 0);" class="elementToProof">
                        I have worked with the Simnibs 3.2.6 headreco to
                        do segmentation and now trying out the charm
                        segmentation routine in the Simnibs 4 Beta
                        release. I have multiple subjects (about 10
                        subjects tried so far) that have looked good in
                        the headreco segmentation in terms of the brain
                        segmentation <span
                          style="background-color:rgb(255, 255,
                          255);display:inline !important">
                          especially </span>using the T1 and T2 but when
                        I try this with the charm segmentation, I run
                        into a problem about 50% of the time. The
                        problem is that the cortical surface is smooth
                        and pushed up right next to the csf boundary
                        layer, but the wm surface looks good. I have
                        attached an image of what is happening to show a
                        better description. I have made sure that the
                        template is aligned by forcing in the code to
                        use the Simnibs 3 warp from MNI2Conform 12
                        parameter alignment, where the previous
                        headmodel seemed to be  accurate. I also try to
                        change some of the settings in the settings.ini
                        file such as all parameters in the segment
                        section and th_initial in the surface section
                        with no avail. As I run the same subject in the
                        Simnibs 3.2.6 the headmodel looks good, I have
                        tried to edit via switching out the brain+csf
                        segmentation from simnibs3 to simnibs 4 and
                        reran the mesh after editing. This works
                        reasonably well but the some of the external to
                        brain structures are affected, plus requires
                        double the runtime to run versions 3 and 4. Have
                        you noticed a problem similar to this and do you
                        have any recommendations to try? Thank you for
                        all your work and I appreciate the research to
                        keep improving this modeling.</div>
                      <div style="font-family: Calibri, Arial,
                        Helvetica, sans-serif; font-size: 12pt; color:
                        rgb(0, 0, 0);" class="elementToProof">
                        <br>
                      </div>
                      <div style="font-family: Calibri, Arial,
                        Helvetica, sans-serif; font-size: 12pt; color:
                        rgb(0, 0, 0);" class="elementToProof">
                        Thank you,</div>
                      <div style="font-family: Calibri, Arial,
                        Helvetica, sans-serif; font-size: 12pt; color:
                        rgb(0, 0, 0);" class="elementToProof">
                        <br>
                      </div>
                      <div style="font-family: Calibri, Arial,
                        Helvetica, sans-serif; font-size: 12pt; color:
                        rgb(0, 0, 0);" class="elementToProof">
                        Joel Upston</div>
                      <div style="font-family: Calibri, Arial,
                        Helvetica, sans-serif; font-size: 12pt; color:
                        rgb(0, 0, 0);" class="elementToProof">
                        <br>
                      </div>
                      <div style="font-family: Calibri, Arial,
                        Helvetica, sans-serif; font-size: 12pt; color:
                        rgb(0, 0, 0);" class="elementToProof">
                        <img alt="Correct_GM_surface_from_headreco.png"
                          src="cid:part1.QCAOTP2r.o28mgR4v@drcmr.dk"
                          class=""><br>
                      </div>
                      <div style="font-family: Calibri, Arial,
                        Helvetica, sans-serif; font-size: 12pt; color:
                        rgb(0, 0, 0);" class="elementToProof">
                        <br>
                      </div>
                      <div style="font-family: Calibri, Arial,
                        Helvetica, sans-serif; font-size: 12pt; color:
                        rgb(0, 0, 0);" class="elementToProof">
                        <img alt="Incorrect_GM_surface_from_charm.png"
                          src="cid:part2.FIFgPlnA.f8JbtZmv@drcmr.dk"
                          class=""><br>
                      </div>
                      <img id="32A1A34F-43FD-42CA-9990-D08828BD128B"
                        alt="Screenshot%20from%202022-08-04%2012-51-01.png"
                        src="cid:part3.1MdYObP0.TFcjq0Pc@drcmr.dk"
                        class=""></div>
                  </blockquote>
                </div>
              </div>
            </blockquote>
          </div>
        </div>
      </div>
    </blockquote>
  </body>
</html>