Hi all! 👋 I'm a PhD student at Johns Hopkins working on medical image processing, specifically super-resolution of MRI.
Super-res is an inverse problem and therefore we have to "make up" information. This is particularly dangerous in medical scenarios, so I'm working on understanding when (if ever!) machine learning-based super-res can safely be applied to MRI with signal processing foundations.
Looking forward to connecting with everyone!
🇳🇱
@sremedios Great to have you on the fediverse.
Yopu might like my pinned post (though probably nothing special for you honestly, but I think its cool):
@freemo Beautiful rendering! Getting that scan must've taken a while in the scanner eh?
@sremedios Wasnt quick thats for sure. But nothing too excessive. Few minutes IIRC maybe 15 minutes if i had to guess, it was a while ago.
@freemo Yeah that's not bad at all!
@sremedios I wish i had access to an MRI just for fun, so many cool things I could potentially do with one... I do have a lab with three different GCMS in it though that i could use for experiments, so at least I got that :)
@sremedios perhaps not as satisfying as processing your own personal data, but still quite interesting!
@freemo I dunno if this will scratch the same itch since it's purely computational, but it looks like you can try playing around with different MR acqs here: https://github.com/mzaiss/MRTwin_pulseq I'm not affiliated with this group and haven't tried it yet, but it seems very fun!