patient name were displayed on the frame so that 
each individual photograph could be clearly 
identifiable. A grid was projected on to the patient's 
back using a slide projector in order to help 
stereoscopic viewing by giving the generally 
featureless back some texture. Colour reversal film 
(slide film) with as ASA rating of 200 was used 
under normal fluorescent lighting conditions. The 
photography was very reliable with clear images 
resulting in all cases. 
3.2 Object Restitution 
Three dimensional shape of the back and positions 
of the markers were measured and recorded using 
a Zeiss C100 Planicomp analytical plotter. In most 
situations the shape was measured as a digital 
elevation model with a 25mm mesh size. The mesh 
data was supplemented with additional points and 
features such as high point of the rib hump and the 
"valley" down the middle of the back. Once the 
height data was collected it was modified according 
to the correction data obtained through the system 
calibration procedure. The measurement phase 
was straight forward but time consuming. This 
would no doubt be improved with the acquisition of 
digital photogrammetric software. 
  
  
  
  
  
Fig. 2 Contour Map of Back 
4. RESULTS 
The initial sets of three stereophotographs have 
been analysed and contour maps created for each 
of twelve patients, for all three sittings. To 
determine a measure of reproducibility these 
contour maps were then subtracted from one 
282 
another (sitting 1 - sitting 2, sitting 1 - sitting 3, 
sitting 2 — sitting 3). The subtracted contour maps 
give a further contour map showing the differences 
between sittings. A perfect match gives a flat 
surface with no contours. 
A number of points (15) were then taken at random 
from each of the subtracted contour maps to give a 
measure of overall point accuracy. The data are 
listed below: 
Event Average (mm) Std Dev (mm) 
Sitting (1-2) 4.1 3.6 
Sitting (1-3) 4.5 4.8 
Sitting (2-3) 3.4 2.9 
Total 4.0 3.9 
From the data we see that as the patient becomes 
more familiar with the positioning frame (sitting (2- 
3) the reproducibility improves to 3.4mm with 
standard deviation 2.9mm. The overall point 
accuracy of the differenced contour maps (or 
repeatability) was 4.0mm with a standard deviation 
of 3.9mm. 
These values appear reasonable when the flexible 
nature of the human body is considered. 
5. CONCLUSIONS 
It would appear that the stereophotogrammetry 
technique developed under this project is 
reproducible and accurate to within 3-4mm. In its 
current state the technique is not yet viable as a 
screening tool but it is very useful as a research 
tool. 
It is proposed to set up two prospective clinical 
trials to assess the change in back shape in 
patients with idiopathic scoliosis undergoing 
surgical intervention. The first is a randomised 
study of braced versus non-braced change in back 
shape in patients who undergo costoplasty as part 
of their surgical stabilisation. The second is to 
document the reduction in rib prominence in 
idiopathic scoliosis patients undergoing surgical 
stabilisation and to compare the amount of 
correction in those patients who undergo 
costoplasty as a part of their surgical procedure 
versus those who have surgical instrumentation 
and fusion only. 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B5. Vienna 1996 
The aul 
financial 
Researc 
D'Amico 
(Editors) 
Spinal d 
Karara, 
Photogr: 
Photogr: 
Slama, 
Photogr: 
Photogr: 
Turner- 
and Jeff 
Back SI 
pp 497-