P3D Re X-Ray photographs in 3D, Re 3D Ultra-sound

[abram klooswyk <abram.klooswyk@xxxxxx>]

Catching up in reading the P3d Digests I just reached the 3D 
X-ray and Ultrasound postings of mid-May. 
Just a few anecdotical additions.

As is clear from Peter Abrahams excerpt, the first stereo X-rays 
were made just a few month after Roentgen had discovered the
rays. Remember that in the late 19th century stereoviews were
normal in the parlors, and probably up to a half of all 
photography was in stereo, so the idea of stereo X-rays was
only self evident.
They indeed played a great role in getting bomb fragments out 
off people in World War I, and this role was activated again 
when fragmentation bombs were used in Vietnam.

The purpose of stereo X-rays of course is localization
in the third dimension, for in any plain X-ray all anatomical
structures are in superimposed on the film. 
But this role nowadays largely has been superseded by imaging 
techniques which provides slices: Computed Tomography (CT), 
Ultrasound (echo) and Magnetic Resonance Imaging (MR or MRI).

However, I remember that some 20 years ago stereo X-rays were 
a standard procedure for a number of difficult areas: 
the skull (especially the orbit, the sella turcica and the 
inner ear) and the neck. 

I know a professor of neurosurgery who had a large 
collection of stereo X-rays of neck fractures and 
dislocations. Not all of them are lethal, and superimposed
left and right sides can only clearly be identified on 
lateral X-rays when seen in stereo. On these films the left 
side of the neck was marked by taping *two* small lead corns 
to that side, and *one* at the right side (no colors on
X-rays, so no red and green mark!). Marking is necessary, 
for with X-rays there is no real pseudoscopy! "Transposing"
lateral stereo X-rays swaps left and right, but you cannot tell
if it has been done when there is no indication how the films 
were taken. You cannot tell whether the side closest to you 
is left or right when the lead corns are omitted.

Hugo de Wijs has been in touch with a professor of
radiology who advocated stereo X-ray in the Netherlands,
in the time before the modern slicing techniques were
available, and at the exhibition of Hugo's stereo work, last
year at the ISU congress in the Netherlands, he had an
old X-ray stereoscope from about 1920 on display, for which I
provided some stereo X-rays.

Another funny thing is that there is a reasonable chance
that stereo X-rays of yourself exist! When at least two
chest X-rays of you have been made, not too far apart
in time, there is a great chance that these differ in
projection and that at least a part of them can be seen in 
stereo. The reason is that the vertical position of the
X-ray tube is chosen by the radiographer, and often at
a retake a month or a year later the tube is positioned 
several cm lower or higher than the first one.
Then there is a vertical parallax, which results in
stereo effects when both films are free viewed after
rotating them over 90 degrees. Admittedly this only
occasionally gives a fine stereo view, and mostly only
ribs stand out in 3D, but sometimes also lung vessels
look like a stereo of a shrub in winter.

The proliferation of X-ray CT investigations on the one hand 
has made stereo X-rays old fashioned, but on the other hand
3D rendering of series of CT images is flourishing, and
again two 3D rendered images differing some 5 degrees in
rotation form a stereopair, and indeed commercial 
radiological imaging work stations can display such
images more or less semi-automatically in stereo, with 
LCD shutter glasses. 

Now for stereo Ultrasound (US), this is possible, and I have
made a few examples. The technique is complex, for normal
ultrasound images again are cuts, slices through an
anatomical structure. As you can imagine, two different
slices from a cake never make a stereopair. 
The first thing is to get an ultrasound image which isn't
a slice but an image of a surface, or a superposition
of a volume as seen from some distance. Second, this
view has to be rotated to get the pair of views required
for stereo. Now the latter technique is implemented in
a variation of the Doppler mode of some machines.
In this mode blood vessels can be seen (because the
blood moves), and vessels in a small volume can be
imaged by rotating the US transducer over a small angle.
Then the software permits to give two different projections of 
the volume, and then the vessel pattern can be seen in stereo. 
But to be honest, the results are not very impressive.

Abram Klooswyk


------------------------------