 |
 |
Xi, I can share with you some of my meager experiences on registration outside of the brain. It all depends on what image features (contrasts) that you can take advantage of and what region is of interest to your application. I suspect skin is not of much interest to your breast MRI application, then trying to match the skin is a moot point and you probably want to mask out the skin first. Then check whether what's left in the image has enough feature for the software algorithm to grab onto. For example, for my prostate application, obviously leg or bladder movement is not of my interest so I mask them out but I leave the rectum wall inside the mask. With that said, the inner edge of the skin to breast tissue is possibly still of importance to your registration. And I also recommend you to try non-linear registration since the breast is not a rigid structure. A lot of trial and error are needed as looks like registration is an ill defined task for your specific application. Regards Gordon On 10/15/2009 03:13 AM, Xi Liang wrote: > Dear all, > > My application is on breast MRI registration, and I have tried FLIRT which removed > the most artifacts. However, there are still many obvious motions from the > subtraction images, like skin motion (very thick skin in subtraction). > > I want to test other software which might do a good job as well. Is there any > recommendation? I know most of softwares are designed for brain. > > Regards, > > Xi Liang >