<html>
<head>
<style>
.hmmessage P
{
margin:0px;
padding:0px
}
body.hmmessage
{
FONT-SIZE: 10pt;
FONT-FAMILY:Tahoma
}
</style>
</head>
<body class='hmmessage'>
Luis and Rupert,<BR>
<BR>
Thank you very much for your valuable input. The frame around the images is intentionally put in so as not to get artifacts when the image processing is done. What I am doing is using these projections to recalculate pseudo-volumetric data akin to CT data. Without this border, there are several artifacts that appear on the final data. I had always (very wrongly) assumed that, if the method does not work with the portal images themselves, it would also not work with the generated data which lacks the "frame". <BR>
<BR>
Rupert, unfortunately, the data I use from my EPIDs is in the form of 8-bit data. I can get unprocessed 16-bit raw data but I have yet to determine how the imaging software post-processes it. The images you saw are actually the version "optimized" for viewing by the imager. Unfortunately, they do not save it as 16-bit. In any case, I will check and see if the method works on my volumetric data. Otherwise, I may ask for more help and post more example images. For now, you have pointed me down the right track and I thank you for it. Best wishes, <BR>
<BR>
Vik<BR><BR>> Date: Wed, 23 Jul 2008 11:33:15 -0400<BR>> From: rupe.brooks@gmail.com<BR>> To: luis.ibanez@kitware.com<BR>> Subject: Re: [Insight-users] Intensity-based registration with low contrast images.<BR>> CC: vikmonster@hotmail.com; insight-users@itk.org<BR>> <BR>> Hi Vikren,<BR>> <BR>> I see Luis just beat me to it! I just ran an ITK based registration<BR>> on these images, using the NormalizedCorrelation Metric, and obtained<BR>> a good registration. As Luis did, I also had to crop off the black<BR>> border.<BR>> <BR>> I got these parameters to map using a translation transform from the<BR>> fixed image to the moving image. 12.3474 -4.05349<BR>> This appears reasonable to me, but i didnt spend too long looking at<BR>> it. I was also able to get a similar result with<BR>> MattesMutualInformation, using 50 bins. (i didnt try other numbers of<BR>> bins)<BR>> <BR>> Because the images are low contrast i would suggest that you be<BR>> careful not to lose data along the way. So, avoid compressed formats<BR>> like JPG, and avoid converting to 8-bit if the original images are in<BR>> 16bit. (Probably you just converted to 8-bit compressed for posting<BR>> anyway.)<BR>> <BR>> Because mutual information uses an evenly spaced set of histogram<BR>> bins, these images might be more difficult to register with MI. It<BR>> worked when i tried it, but i am almost certain its just registering<BR>> the fairly bright head, to the dark background, and not actually using<BR>> any of the detail inside. In any case, if these images are<BR>> representative of your problem, i think MI is unnecessary.<BR>> <BR>> Cheers,<BR>> Rupert B.<BR>> <BR>> On Wed, Jul 23, 2008 at 11:08 AM, Luis Ibanez <luis.ibanez@kitware.com> wrote:<BR>> ><BR>> > Hi Vikren,<BR>> ><BR>> > Thanks for the clear description of your problem and for posting the<BR>> > images in an accessible place.<BR>> ><BR>> > The problem in your case, is not the low contrast of the images<BR>> > (although that certainly doesn't help...). The main problem is<BR>> > related to how these images have been composed.<BR>> ><BR>> > --<BR>> ><BR>> > I just ran on them the example ImageRegistration9<BR>> > (without modifications), an obtained a reasonable registration.<BR>> ><BR>> > Looking closer at your images, the problem is that you seem to have<BR>> > copy/pasted an image on top of a larger canvas.<BR>> ><BR>> > That is, your images are of size 512x512, with a background of value 0.<BR>> > Insight this region, there is a rectangle from (25,23) to (488,488) at<BR>> > a intensity level close to 20 at the top, and with a gradient that<BR>> > grows upto 45 at the bottom of the image.<BR>> ><BR>> > Inside that region you have the projection of the head.<BR>> ><BR>> ><BR>> > The rectangular region on the background is preventing the registration<BR>> > from being driven by the data from pixels inside the head.<BR>> ><BR>> ><BR>> > You can get rid of this problem in several ways.<BR>> ><BR>> ><BR>> > a) you can crop the images in order to contain only the<BR>> > region (24,24) to (488,488) before you feed the images<BR>> > into the registration.<BR>> ><BR>> > or<BR>> ><BR>> ><BR>> > b) inside the registration you can set the FixedImageRegion<BR>> > to the region (24,24) to (488,488).<BR>> ><BR>> ><BR>> ><BR>> > Regards,<BR>> ><BR>> ><BR>> > Luis<BR>> ><BR>> > ----------------------<BR>> > Vikren Sarkar wrote:<BR>> >><BR>> >> Greetings,<BR>> >> I am trying to see if it is possible to perform intensity-based<BR>> >> registration on some very low contrast images that I have. I tried posting<BR>> >> this question a few days ago but realize there were some problems with the<BR>> >> website I was using. Hopefully photobucket will be better. The following two<BR>> >> images were acquired using an electronic portal imaging device. The only<BR>> >> difference is that one image is translated compared to the other.<BR>> >> File 1 : http://i336.photobucket.com/albums/n333/poohvee/fixed.jpg<BR>> >> File 2 : http://i336.photobucket.com/albums/n333/poohvee/moving.jpg<BR>> >> Being new to ITK, I first started with several of the example<BR>> >> registration codes but none seem to work with the low contrast level. Luis<BR>> >> also suggested I try to run the images through edge detection. I tried that<BR>> >> approach as well as thresholding before registering but could not get any<BR>> >> results with any of the example registration codes. Finally, I did try the<BR>> >> example code using mutual information (example 2 in image registration) and<BR>> >> it did not work either.<BR>> >> Does anyone have any other suggestion I can use to artificially increase<BR>> >> the contrast before registering or can someone suggest a metric that would<BR>> >> be better suited for the task? The final aim would be to try to use this<BR>> >> with 3D datasets but I would like to start with 2D. Thank you.<BR>> >><BR>> >> Vik<BR>> >><BR>> >><BR>> >> ------------------------------------------------------------------------<BR>> >><BR>> >> _______________________________________________<BR>> >> Insight-users mailing list<BR>> >> Insight-users@itk.org<BR>> >> http://www.itk.org/mailman/listinfo/insight-users<BR>> ><BR>> > _______________________________________________<BR>> > Insight-users mailing list<BR>> > Insight-users@itk.org<BR>> > http://www.itk.org/mailman/listinfo/insight-users<BR>> ><BR>> <BR>> <BR>> <BR>> -- <BR>> --------------------------------------------------------------<BR>> Rupert Brooks<BR>> McGill Centre for Intelligent Machines (www.cim.mcgill.ca)<BR>> Ph.D Student, Electrical and Computer Engineering<BR>> http://www.cyberus.ca/~rbrooks<BR><BR><br /><hr />Time for vacation? WIN what you need. <a href='http://www.gowindowslive.com/summergiveaway/?ocid=tag_jlyhm' target='_new'>Enter Now!</a></body>
</html>