NavList:

   

Reply

On 23 Feb 2005 at 14:44, Frank Reed wrote:

> Ken M you wrote:

> "In video microscopy it's common to go beyond the diffraction limit  and
> pick out features (but not "detail") that should be too small to see.  This
> works because of motion (video microscopy provides 30  frames/sec).
> Although on a still photo one would not be able to distinguish  features
> from artifacts, persistance and motion give additional clues that  allow
> one to discriminate. Artifacts don't obey Newton's laws, for  example."
>
> Are you going beyond the diffraction limit? Or beyond the noise  limit?

The diffraction limit for light microscopy is about 230 nm (0.61 *
wavelength / numerical aperture). With video-enhanced differential
contrast interference microscopy, it's possible to see structures that are
only 20 nm in diameter when measured using electron microscopy. This is
done by using an electronic background adjustment to increase the contrast
and by sharpening blurry smudges. Some of those smudges are artifacts, but
the ones that do things in an expect manner (e.g. a kinesin motor crawling
along a microtubule) are considered real.

> Is
> this just the vision system's ability to average out  noise in time-varying
> inputs?

That's a part of it, but not the whole story.

> I suppose everyone is familiar with the fact that  a single frame
> of video can be very grainy and look completely smooth when  displayed as a
> video stream. Capture a single frame of tyical television and the  effect
> is dramatic. >From a sextant user's point of view, maybe this means we
> should be happy our hands shake.

Hmmm...it sure seems like it would be easier if my hands were steady.

> And:
> "Somehow the brain puts all this detail into
> the moving image that's not  really there (except that it is there--i.e. if
> you're familiar with  histological sections of the same tissue, the MRI
> detail matches the genuine  anatomical detail). I have no idea how this
> works, but it is amazing to  witness."
>
> Fascinating. You're talking about viewing the "slices" like a fast
> slideshow, right?

Yes, some MRI machines have a trackball for moving in or out (flipping
through the slices) so that you can do it slowly or quickly. My
recollection is that there is an optimal speed for seeing detail in the
images but it's been quite a while since I've had access to a machine so I
may be mistaken.

> Does familiarity with the tissue structure help here, or
> are  you saying that it confirms what you see whether you're familiar or
> not?

I don't know. Certainly you see detail that is not seen in the static
images even when you don't know what the tissue should look like, but I
haven't subsequently compared those observations to histological sections
to test whether that detail was accurate. Only for tissues where I knew
the anatomy could I say that the extra detail appeared as it should.

It would be interesting to take a series of images from a stop-motion
sequence (perhaps using line-art animation) and subtract information and
then see if the animated sequence allows the brain to reconstruct the
objects. It may be that the MRI images contain all the "missing"
information but we just can't process it properly with a static image.

Ken Muldrew.

   

Reply