On 6-2-2010 18:56, Sabrina Kilian wrote:
> Darren New wrote:
>> Sabrina Kilian wrote:
>>> Technical reason that the camera could not get closer. The Magnetom Trio
>>> (name on the machine) is a 3 Tesla MRI. A digital camera, or anything
>>> ferrous, would not stand a chance near that.
>> Well, the dude on the outside is looking at *something*.  I'm assuming
>> there's some electronics-based display in the control room or something.
>>
> 
> Having never been in one of those specific machines, I can't say for
> sure. Most of the displays, however, don't do all of the fancy graphics
> and colors as the machine is running. Secondly, to show all of the scan,
> they would have to introduce a radiologist or rad-technician, as most
> hospitals would not let a doctor of mathematics operate an MRI (he could
> be trained as a tech, I suppose). Less snappy for TV. Also, those
> multiple displays in the control area are not usually for showing the
> MRI data. One or two would display the rather ancient UI, with large
> numbers displaying how far the bed is from it's rest position and other
> diagnostic data from the machine itself.
> 
> I don't have any MRI technicians I can email about how fast the displays
> in the control room are updated. I know that for sonograms, the display
> is generally just for sighting, even the false color is just a quick
> render to aid the technician in getting the image they want. The print,
> film, file, whatever, version is what is used for diagnosis.
> 
> Next time I get near an fMRI, I will pester the techs about how fast
> they can shift what is on the control display, and whether it can keep
> up with the scanner in real-time.

The image is generally shown in the screen before the next image is 
taken. Unless a sequence is taken at once. It can be analyzed in full 
detail 'immediately' in 2D on-line. 'Immediately' is generally a 
significant number of seconds. 3D reconstruction is off-line but faster 
than it would take POV to render.

However, that is not the point here. They take good resolution slices of 
the whole brain for research purposes. If they really wanted to a 
predictor they would use a much faster protocol to get only the data in 
the area that they now know is relevant. A dedicated prediction machine 
would take less than a second to analyse that small part of the brain.

Post a reply to this message