[Mristudio-users] About loaded ROI into my Anatomy Image

Fri Jul 22 19:08:24 EDT 2011

Hi Jacquie,

Defining a brain structure or ROI is usually performed by manual drawing.
The manual definition is usually treated as a gold standard (if it is done
by an experienced person). However, as you pointed out, it has several
problems; it requires training, time consuming, poor reproducibility, etc.
Therefore, more automated approaches are much needed.  The issue you raised,
low resolution or poor contrast, is a more fundamental issue, because it
affects both manual and automated approaches. For example, if the resolution
is poor to reliably delineate a structure manually, the chance is high that
the automated method is not reliable either.

When you choose to use an automated method, you have to tell the computer
which structure you want to define, like hippocampus. Here, there are
several points I want to make;

1) There are often multiple ways to define a structure of interest. For
example, different anatomist may define the hippocampus in slightly
different ways. In literature, the size of the hippocampus could be
different as much as twice. In this context, "well-known ROI" means
"well-described definition  by a famous neuroanatomist about how to define
the hippocampus". You may find that there are surprisingly a small number of
publications for this type of topics. For the hippocampus, I recommend you
to go over some work by European consortium by Dr. Giovanni B. Frisoni
2) Some may want only a subregion of the hippocampus, like CA1, CA2, etc. As
you try to define more and more detailed structures, you find that the
smallest structure you can possibly define is the image pixels. Namely, you
can theoretically define corresponding pixel coordinates across subjects.
This is the idea of the voxel-based analysis. In this case, your ROI is one
pixel. One of the most "well-known ROIs" in this case is the MNI
coordinates. Instead of saying, "an ROI in the hippocampus", you can say,
"MNI coordinate X, Y, Z".
3) However, usually, when one says, "ROI", that is not a single pixel, but a
cluster of pixels that contain an anatomical representation, such as "the
hippocampus". In this case, computer algorithms often require a prior
definition. Many programs use a prior probabilistic map based on
manual delineation. I believe a program like FreeSurfer comes with prior
probabilistic maps. In this case, these maps serve as "well-known ROI".
4) Some other programs (like our MriStudio) use a pre-segmented atlas. As I
mentioned above (#1), there is some degree of arbitrariness when we define a
structure. However, once a structure is defined in an atlas, we can claim
that we can apply a consistent definition to all images. There are widely
used atlases for cortical ROIs (such as AAL and PickAtlas). The MriStudio
program family (RoiEditor) comes with a pre-segmented atlas called, "white
matter parcellation map (WMPM). This is a bit misnomer, because it also
contains parcellation of the gray matter. I'm not sure if we can call it
"well-known", but it is an option.
5) The approach I described in #4 is basically the same as the voxel-based
approach (#2) because you need to warp the atlas (e.g. WMPM) to other
subjects, which requires voxel-to-voxel mapping between the two images. The
voxel-to-voxel mapping is performed independent of any structural
definitions. This means, once voxel-to-voxel mapping (called image
normalization or transformation) is completed, you can define any arbitrary
structure in one brain and this definition is automatically transferred to
the other. Therefore, you can define your own ROI (do not have to use the
WMPM we provide). In this context, whether the ROI is "well-known" or not
seems not a critical issue to me. What is more critical is if the
voxel-to-voxel mapping is accurate or not.

I hope you find this helpful.

Susumu

On Wed, Jul 20, 2011 at 4:33 PM, Jacquie Hodge <
Jacquie.Hodge at albertahealthservices.ca> wrote:

> Hello All,
>
> I have been going through older emails to try and answer my own questions
> and came across this email. Can someone please share with me a "well known
> Atlas" and maybe elaborate a bit on the uses for these, how it would help me
> and what I do with it?
>
> Thank you in advance :)
>
> Jacquie
>
> Jacquie Hodge MSc candidate in Neuroscience, BSc
> Neuroscience, University of Calgary
> Research Assistant, Calgary Pediatric Stroke Program
> Alberta Children's Hospital
> 2888 Shaganappi Trail NW, Calgary, AB, Canada T3B 6A8
> Phone: 403-955-7733 Fax: 403-955-2922
> jacquie.hodge at albertahealthservices.ca
> Program website: www.perinatalstroke.com
> ________________________________________
> From: mristudio-users-bounces at mristudio.org [
> mristudio-users-bounces at mristudio.org] On Behalf Of [Seongjin] [
> choisj70 at gmail.com]
> Sent: February 28, 2011 10:15 AM
> To: DTI Studio, ROI Editor, DiffeoMap Questions/Support
> Subject: Re: [Mristudio-users] About loaded ROI into my Anatomy Image
>
> I believe that you should do spatial normalization of your data to a well
> known ATLAS first.
>
> SC
>
>
>
> On Mon, Feb 28, 2011 at 8:15 AM, Yibao Wang <cmuwyb at gmail.com<mailto:
> cmuwyb at gmail.com>> wrote:
>
>
> Is there a method to load the well-known ROI into my Anatomy Image?
> I explain my question in detail,
> for example, the are lots of well-known ROI such as insular, hippocampus,
> it was diffcult to draw it in the anatomy image, sometimes because its poor
> resoultion.
> Could we loaded those ROI into my Image?
> I wonder is there any software could analysize the well-konwn interested
> ROI from the poor resolution anatomy image.
>
>
> _______________________________________________
> mristudio-users mailing list
> mristudio-users at mristudio.org<mailto:mristudio-users at mristudio.org>
> http://lists.mristudio.org/mailman/listinfo/
> Unsubscribe, send a blank email to:
> mristudio-users-unsubscribe at mristudio.org<mailto:
> mristudio-users-unsubscribe at mristudio.org>
>
>
>
>
>
>        <<
>                ATT00001..c     (0.3KB)
>
>                (0.3KB)
>        >>
>
> This message and any attached documents are only for the use of the
> intended recipient(s), are confidential and may contain privileged
> information. Any unauthorized review, use, retransmission, or other
> disclosure is strictly prohibited. If you have received this message in
> error, please notify the sender immediately, and then delete the original
> message. Thank you.
>
> _______________________________________________
> mristudio-users mailing list
> mristudio-users at mristudio.org
> http://lists.mristudio.org/mailman/listinfo/
> Unsubscribe, send a blank email to:
> mristudio-users-unsubscribe at mristudio.org
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.mristudio.org/pipermail/mristudio-users/attachments/20110722/bd560355/attachment.html 