I talked with Xin and confirmed that the LDDMM-AIR combined transformation matrix has the same format as K/H maps. They are raw files of Vector x X x Y x Z. <br><br><div class="gmail_quote">On Mon, Feb 1, 2010 at 10:22 AM, Xin Li <span dir="ltr"><<a href="mailto:xli16@jhmi.edu">xli16@jhmi.edu</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">Hello Yi,<br>
<br>
As Susumu said, transformation matrices contain vectors at each coordinate, indicating where is the new location. The structure is Vector (x, y, z) x X-dimension x Y-dimension x Z-dimension. You may write matlab code to extract the vector information.<br>
<br>
<br>
Xin<br>
<div class="im"><br>
<br>
<br>
<br>
<br>
----- Original Message -----<br>
From: susumu mori <<a href="mailto:susumu@mri.jhu.edu">susumu@mri.jhu.edu</a>><br>
Date: Saturday, January 30, 2010 8:29 am<br>
Subject: Re: [Mristudio-users] apply transformation to landmarkers & command file<br>
</div><div class="im">To: Yi Jiang <<a href="mailto:yj3@duke.edu">yj3@duke.edu</a>>, "DTI Studio, ROI Editor, Landmarker Questions/Support" <<a href="mailto:mristudio-users@mristudio.org">mristudio-users@mristudio.org</a>>, Xin Li <<a href="mailto:xli16@jhmi.edu">xli16@jhmi.edu</a>>, Jiangyang Zhang <<a href="mailto:jzhang3@jhmi.edu">jzhang3@jhmi.edu</a>><br>
<br>
<br>
> Hi Yi,<br>
><br>
> On Fri, Jan 29, 2010 at 5:25 PM, Yi Jiang <<a href="mailto:yj3@duke.edu">yj3@duke.edu</a>> wrote:<br>
><br>
> > Hello, dear all,<br>
> ><br>
> ><br>
> ><br>
> > I have two questions and am wondering if someone can help.<br>
> ><br>
> ><br>
> ><br>
</div>> > 1. I have already had a transformation matrix to map the subject<br>
<div class="im">> image<br>
> > onto the template image, which was generated by combining AIR<br>
> rigid, AIR<br>
> > affine, and LDDMM matrix. I want to define some landmarkers on<br>
> the subject<br>
> > image, and apply the transformation matrix to the landmarkers<br>
> (to see their<br>
> > corresponding location on the template image). How can I do that?<br>
> ><br>
> ><br>
> I think this is exactly what the K and H maps are. I need<br>
> confirmation from<br>
> Xin, but in my understanding, these transformation matrices contain vectors<br>
> at each coordinate, indicating where is the new location. If I'm not<br>
> mistaken, it is Vector (x, y, z) x X-dimension x Y-dimension x Z-dimension.<br>
> There must be a way to extarct the "Vector" information at the specified<br>
> coordinate.<br>
><br>
> ><br>
> ><br>
</div>> > 1. I would like to use the “load command file” function. I will<br>
<div><div></div><div class="h5">> do AIR<br>
> > linear rigid transformation, AIR linear affine transformation,<br>
> image format<br>
> > change to byte, automatic intensity correction by histogram, and<br>
> 2-channel<br>
> > LDDMM. Is it possible to put all these into one command file? If<br>
> yes, can<br>
> > anyone give me an example file?<br>
> ><br>
> > I believe the command file is only for the functions within the<br>
> same box,<br>
> including byte-integer-float conversion, axial-sagittal-coronal conversion,<br>
> resampling, etc. They don't include transformation functions. We are<br>
> working<br>
> on automation, which will take several months. Meanwhile, I believe<br>
> it is<br>
> good idea to use the current way, which forces you to check the<br>
> result after<br>
> each step. Transformation sometimes does silly things. Some images<br>
> may have<br>
> quality problems. You may want compare different methods (e.g. Mutual<br>
> Info<br>
> linear transformation vs AIR, intensity correction methods, contrast<br>
> of<br>
> choice). I think it's very important to get a feel about your data<br>
> and each<br>
> transformation step. Also, depending on pathology, the transformation<br>
> strategy may change. Some disease groups have severe intensity<br>
> changes (e.g.<br>
> stroke), some have severely enlarged ventricles. You may find the same<br>
> methods may work for one but not the other. Automation is still a difficult<br>
> matter for transformation-based analysis.<br>
><br>
><br>
> ><br>
> ><br>
> > Thank you very much!<br>
> ><br>
> ><br>
> ><br>
> > Have a nice weekend!<br>
> ><br>
> ><br>
> ><br>
> > Best<br>
> ><br>
> > Yi<br>
> ><br>
> ><br>
> ><br>
> ><br>
> ><br>
> > _______________________________________________<br>
> > Mristudio-users mailing list<br>
> > <a href="mailto:Mristudio-users@mristudio.org">Mristudio-users@mristudio.org</a><br>
> ><br>
</div></div><div><div></div><div class="h5">> > Unsubscribe, send a blank email to:<br>
> > <a href="mailto:Mristudio-users-unsubscribe@mristudio.org">Mristudio-users-unsubscribe@mristudio.org</a><br>
> ><br>
> ><br>
<br>
</div></div></blockquote></div><br>