<html xmlns:v="urn:schemas-microsoft-com:vml" xmlns:o="urn:schemas-microsoft-com:office:office" xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:m="http://schemas.microsoft.com/office/2004/12/omml" xmlns="http://www.w3.org/TR/REC-html40">
<head>
<meta http-equiv=Content-Type content="text/html; charset=utf-8">
<meta name=Generator content="Microsoft Word 12 (filtered medium)">
<style>
<!--
/* Font Definitions */
@font-face
{font-family:"Cambria Math";
panose-1:2 4 5 3 5 4 6 3 2 4;}
@font-face
{font-family:Calibri;
panose-1:2 15 5 2 2 2 4 3 2 4;}
@font-face
{font-family:Tahoma;
panose-1:2 11 6 4 3 5 4 4 2 4;}
/* Style Definitions */
p.MsoNormal, li.MsoNormal, div.MsoNormal
{margin:0cm;
margin-bottom:.0001pt;
font-size:12.0pt;
font-family:"Times New Roman","serif";}
a:link, span.MsoHyperlink
{mso-style-priority:99;
color:blue;
text-decoration:underline;}
a:visited, span.MsoHyperlinkFollowed
{mso-style-priority:99;
color:purple;
text-decoration:underline;}
span.EmailStyle17
{mso-style-type:personal-reply;
font-family:"Calibri","sans-serif";
color:#1F497D;}
.MsoChpDefault
{mso-style-type:export-only;}
@page WordSection1
{size:612.0pt 792.0pt;
margin:72.0pt 72.0pt 72.0pt 72.0pt;}
div.WordSection1
{page:WordSection1;}
-->
</style>
<!--[if gte mso 9]><xml>
<o:shapedefaults v:ext="edit" spidmax="1026" />
</xml><![endif]--><!--[if gte mso 9]><xml>
<o:shapelayout v:ext="edit">
<o:idmap v:ext="edit" data="1" />
</o:shapelayout></xml><![endif]-->
</head>
<body lang=EN-US link=blue vlink=purple>
<div class=WordSection1>
<p class=MsoNormal><span style='font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'>Hi Kerstin,<o:p></o:p></span></p>
<p class=MsoNormal><span style='font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'><o:p> </o:p></span></p>
<p class=MsoNormal><span style='font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'>Based on my experience (simulations actually), I found that the
residuals from CSD generated signals *<b>don’t</b>* allow accurate variance
estimates with residual bootstrapping.<o:p></o:p></span></p>
<p class=MsoNormal><span style='font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'><o:p> </o:p></span></p>
<p class=MsoNormal><span style='font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'>Calculating the residuals from the least square SH fit, however,
produced very accurate variance estimates (for details, see <a
href="http://www3.interscience.wiley.com/journal/123476777/abstract">http://www3.interscience.wiley.com/journal/123476777/abstract</a>).<o:p></o:p></span></p>
<p class=MsoNormal><span style='font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'><o:p> </o:p></span></p>
<p class=MsoNormal><span style='font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'>If you are interested, I can provide you with Matlab code that
allows you to use this method.<o:p></o:p></span></p>
<p class=MsoNormal><span style='font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'><o:p> </o:p></span></p>
<p class=MsoNormal><span style='font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'>The problem with using the signal from CSD FOD’s is that it is
not clear how to correct for leverage (equation 8 in the mentioned paper),
since the CSD operation is iterative and uses the non-negativity constraint. If
you just leave out the leverage term in the residual bootstrapping procedure, then
your variance estimates will be biased to a great extent. If I just use the
leverage values from the last iteration, the bias becomes less but is still
much larger then when residual bootstrapping with the SH model.<o:p></o:p></span></p>
<p class=MsoNormal><span style='font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'><o:p> </o:p></span></p>
<p class=MsoNormal><span style='font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'>Cheers,<o:p></o:p></span></p>
<p class=MsoNormal><span style='font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'>Ben<b><o:p></o:p></b></span></p>
<p class=MsoNormal><span style='font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'><o:p> </o:p></span></p>
<p class=MsoNormal><span style='font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'><o:p> </o:p></span></p>
<p class=MsoNormal><span style='font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'><o:p> </o:p></span></p>
<div style='border:none;border-top:solid #B5C4DF 1.0pt;padding:3.0pt 0cm 0cm 0cm'>
<p class=MsoNormal><b><span style='font-size:10.0pt;font-family:"Tahoma","sans-serif"'>From:</span></b><span
style='font-size:10.0pt;font-family:"Tahoma","sans-serif"'> mrtrix-discussion-bounces@public.nitrc.org
[mailto:mrtrix-discussion-bounces@public.nitrc.org] <b>On Behalf Of </b>Donald
Tournier<br>
<b>Sent:</b> donderdag 10 juni 2010 4:01<br>
<b>To:</b> k.pannek1@uq.edu.au<br>
<b>Cc:</b> mrtrix discussion<br>
<b>Subject:</b> Re: [Mrtrix-discussion] calculating recovered HARDI / residuals<o:p></o:p></span></p>
</div>
<p class=MsoNormal><o:p> </o:p></p>
<p class=MsoNormal>Hi Kerstin,<o:p></o:p></p>
<div>
<p class=MsoNormal><o:p> </o:p></p>
</div>
<div>
<p class=MsoNormal>There is no code to do that currently, but it really
shouldn't be too hard. The simplest thing to do would be to copy and modify
cmd/sdeconv.cpp (the simple linear spherical deconvolution), to produce the
modeled DW signals. You'll need to invert the operation, so that the SH
coefficients file is the input, and the DWI file is the output. You can remove
a bunch of options that you won't need (lmax, mask, filter & normalise) and
all the associated code that depends on those options. You'll be determining
lmax from the CSD file, using DWI::SH::LforN (SH.dim(3)), and instead of
converting from amplitudes to SH, you'll be doing the opposite operation - at
line 213, use SHT.SH2A (sigs, res). There's obviously quite a lot more work
to it, but the rest should be *relatively* trivial...<o:p></o:p></p>
</div>
<div>
<p class=MsoNormal><o:p> </o:p></p>
</div>
<div>
<p class=MsoNormal>Get in touch if you get stumped...<o:p></o:p></p>
</div>
<div>
<p class=MsoNormal>Cheers, and good luck!<o:p></o:p></p>
</div>
<div>
<p class=MsoNormal><o:p> </o:p></p>
</div>
<div>
<p class=MsoNormal>Donald.<o:p></o:p></p>
</div>
<div>
<p class=MsoNormal style='margin-bottom:12.0pt'><o:p> </o:p></p>
<div>
<p class=MsoNormal>On 10 June 2010 11:04, Kerstin Pannek <<a
href="mailto:k.pannek1@uq.edu.au">k.pannek1@uq.edu.au</a>> wrote:<o:p></o:p></p>
<p class=MsoNormal>Hi<br>
<br>
I would like to calculate the residuals of my HARDI data for bootstrapping.<br>
Is there a way in mrtrix to recover the HARDI signal from the csd, or could you
let me know how to bring csd, response and gradient directions together to get
the HARDI signal?<br>
<br>
Thanks<br>
Kerstin<br>
_______________________________________________<br>
Mrtrix-discussion mailing list<br>
<a href="mailto:Mrtrix-discussion@www.nitrc.org" target="_blank">Mrtrix-discussion@www.nitrc.org</a><br>
<a href="http://www.nitrc.org/mailman/listinfo/mrtrix-discussion"
target="_blank">http://www.nitrc.org/mailman/listinfo/mrtrix-discussion</a><o:p></o:p></p>
</div>
<p class=MsoNormal><br>
<br clear=all>
<br>
-- <br>
Jacques-Donald Tournier (PhD)<br>
Brain Research Institute, Melbourne, Australia<br>
Tel: +61 (0)3 9496 4078<o:p></o:p></p>
</div>
</div>
</body>
</html>