e contours
same skin
composite
composite
10 volume
6. FUTURE WORKS
In this study, we had planned to integrate external face model of
the patient with internal brain models. We have written a pho-
togrammetric software for external face reconstruction with real
world coordinates. But we couldn't test it within a real medical
application, because of our equipmental limitations. But in the
shortest time when we overcome these problems, we will test it
and integrate outer model with internal medical data. By using
control points we have taken both digital photogrammetric
images and CT and MR images for this study. But we couldn't
match outer face points because of the speed limitations as
mentioned in section 4. On the other side, CT and MR scanners
at our university hospital isn't connected to external computers.
The images are either being published on films or recorded to a
Magnetic Optical Disk (MO disk). We have recorded images
MO disk but we couldn't find any driver to transfer the data to
our computer environment. We have published the images on to
films. And tried to scan films with scanners. But the result
images were too noisy and many important details were lost
during scanning. But in the scanned CT image, our markers
were seen with some bone structures. By using these marker
images we have tested our program's coordinate precision. And
as preliminary results we achieved +.4 mm coordinate preci-
sion, with 5mm. slice thickness and with the cylindrical lead
markers with 10 mm height and 10 mm radius. Here again we
saw that, lead markers caused very great image artifacts be-
cause of X-ray reflections.
In the future works, we will repeat data acquisition procedure
and test our software.
7. CONCLUSION
With 3D medical reconstruction techniques, diagnoses of ab-
normalities is very easy. According to the 3D analysis of the
image slices, treatment plans can be prepared more precise than
classical 2D images. By integrating the medical data with ex-
ternal body surface of the patient, by using photogrammetric
techniques, it will be possible to define the exact location of
tumours from the real reference points on the outer body sur-
face. Thus, for surgical operations with local techniques would
be easier. It would be possible to track the changes of the ab-
normalities by the time. For this purpose, registration of tempo-
ral medical images taken at different times of treatment duration
would be possible with real coordinates of the some reference
points. The artificial neural networks can be integrated to such a
system for automatic diagnosis. And thus, it would be possible
faster diagnosis of cancer.
8. REFERENCES
Gonzales, R., C., 1987. Digital Image Processing. Addison
Wesley Publ.Comp, USA.
http://www.cc.gatech.edu/scivis/tutorial/tutorial. html (accessed
24 March 2001)
Lorensen, W.,E., and Cline, H.,E., 1987, Marching Cubes: A
High Resolution 3D Surface Construction Algorithm. Compu-
ter Graphics, 21(3), pp 163-169.
Ney, D., et.al., 1990, Volume Rendering of Computed Tomo-
graphy Data: Principles and Techniques. IEEE Computer Gra-
phics & Applications, pp 24-32.
Schroeder ,W., et al, 1998, The VisualizationToolkit An
Object Oriented Approach to 3D Graphics. Prentice Hall PTR,
New Jersey.
Teuber, J., 1993, Digital Image Processing, Prentice Hall, UK.
—255—
re mmm
