[Insight-users] Segmentation Question : Breast MRI

[Luis Ibanez]

Hi Neha,

It is not wise to process a medical image before getting familiar
with the acquisition process that was used for generating it. You
may have to do some research with whoever provided the images to
you.

The first fundamental thing to figure out is whether the image was
acquired with a contrast agent or not.

You will understand better what I meant by 'images acquired in well
controlled conditions' if you look into the acquisition process for
breast MRI. Please look at:

    http://imaginis . com/breasthealth/mri . asp#how

Region growing is not probably what you need for this application.
Breast abnormalities may be very small and with non-convex shapes.

You may want to figure out first the acquisition conditions of the
images, and get together with an clinical expert that would be in
the position of showing you examples of commonly found abnormalities.


Please look at the following link for some illustrative examples

    http://imaginis . com/breasthealth/mri_casestudy . asp



   Regards,


     Luis


-----------------
Neha D wrote:
> Hi Luis,
>  
> Thanks for a very detailed reply. Currently, with the MRIs that i have, 
> I am not sure if they are the images of normal breast or breast with 
> abnormalities.Also, I have no absolutely no idea about if those images 
> are before or after injection of gadolinium contrast agent.
> Currently, I am doing a research if i can use ITK for making this clear 
> if there is any abnormaility in the breast MRI.
>  
> So, without knowing this can I say, ITK facilitate the 
> delineation/quantification of abnormailities? How? Can I proceed with 
> any of the Region Growing Segmentatikn Algo. in ITK?
>  
> Also, I did not understand what you meant by "images acquired in well 
> controled conditions". Can you please explain that to me.
> 
> Thanking you in advance,
> Neha
> 
>