Slashdot Mirror
A 3-D View of the Brain
Jamie found a nifty story about Researchers at Thomas Jefferson University Hospital coming up with new 3D Brain Imaging Software. The interesting bit is that it merges data from MRIs as well as various other types of brain scans to create a single visualization for your noodle.
As well as TFA there's a 'Multimedia' link which give much more info - as well as having some pretty pictures.
init 11 - for when you need that edge.
http://www.technologyreview.com/player/07/08/06Sau ser/1.aspx
Hey! That's my sig you're smoking there!
All MRI machines can do it all; they are just different programs you feed the machines to get different images. Unfortunately, the images at the end still have to be lined up. This is typically done by allowing the brains to rotate in 3 dimensions until the registration maximizes some function; for example, the mutual information between the two images. See the package fsl (http://www.fmrib.ox.ac.uk/fsl/) for some great fMRI analysis tools, including FLIRT for aligning brains (of multiple patients, or one patient's fMRI scan to MRI scan).
I work within the field of medical imaging, and this is nothing new. People have been doing image fusion with images from different image modalities for over a decade. There are lots of products like this one, some even open source and with more impressive screenshots. Why is this particular product, which is not even named or referenced, featured? If you want to see impressive open source work within the medical world, check out ITK and VTK (http://www.vtk.org/ and http://www.itk.org/). Now that is really cutting edge work done with free software.
they're relatively easy to match up!
Not always true. A GRE, SE, or FLAIR image sequence for anatomy will not line up well with the EPI sequence of fMRI due to B field non-linearities and shift even if the patient doesn't move. The nice thing, though, is that unless there is surgery and deformation due to swelling, tissue void, or skull shifting, the skull shape stays constant and one can use it as a rigid body for starting the registration.
There are some software programs to attempt it but it still comes down to an expert setting some landmarks and using some validated software. Sorry, but the "free software" world is behind the commercial software and I see the gap widening.
Bill
Fortunately, most brains (unlike arms and legs) aren't in the habit of moving around a lot during an MRI scan, so they're relatively easy to match up!
You've obviously never spent an hour inside one of those machines. I used to do research in an fMRI lab and even something like post nasal drip eventually makes you swallow just to keep breathing and the slight movement pushes your head into a new pixel lattice so when you subtract the images you just see gray everywhere.
> While it is good to see more talented people working in the medical visualization space,
> this is not really a new thing. Image Fusion [wikipedia.org] has been around for a while now
> but it has not yet become a mainstream technique.
It is very mainstream for PET/CT fusion. Many manufacturers make combo PET/CT machines for just this purpose since the acquisitions are done at the same time, they align very closely and little if any rotation/translation has to be done for a good volume match.