Luis Torres

Thanks Nick. I was basing my question on this publication in which they develope a single optimization framework for groupwise registration. https://www.nature.com/articles/s41598-018-31474-7 I guess my next question is does antsMotionCorrection register each frame to the reference frame space directly, or does it register to to the frame next to it and then stack the transforms in a daisychainlike fashion to move the current frame to the reference frame? Thanks again, Luis

Hi all, I would like to register a ridiculous number of images in a timeseries (say, 600ish 3D volumes, whatever realistically fits in memory). Ideally I would like to register in a groupwise fashion as opposed to 600 pairwise registrations. Is this possible with ants? I guess what I am looking for is whether or not ants can use a groupwise similarity metric? If not, is there some extra documentation on the antsMotionCorrection script? I've been through the --help and the antsDoc, but I'm still uncertain...

Thanks Nick. After reading the literature (and your cartoon explanation above!) I have a much better understanding of how the correction works. Much appreciated. Also, learning how to use ANTs has been a great experience, so thank you guys! Luis

Thanks Nick, this clears some things up. The reason I brought this up was because I'm seeing hot spots(cold spots?) in the lung cavities in my bias field images if I include them in the mask (attached). I expected to see a somewhat uniform decrease in signal towards the middle of the subject due to the surface coils being placed at the anterior and posterior of the subject, which is what I see when I exclude the lungs from the mask(also attached). I suspect the hot spots aren't completely accurate,...

Hi Nick, I'm currently using N4BiasFieldCorrection to correct for loss of signal in 3D MRI chest images. I mask out the lungs/background (Otsu works terribly with these images). I was hoping you could help solidify my understanding of how N4 works. In order to obtain an accurate bias field, should I be including or excluding the lungs, since they have a much lower signal intensity than the rest of the body? It seems to me that N4 will overcorrect the lung regions if they are included in the mask....

This seems to be the latest update I can find on gpu support, Is there any new updates on this or an expected timeline? Thanks! Luis