You are using an outdated browser that does not fully support the intranda viewer.
As a result, some pages may not be displayed correctly.

We recommend you use one of the following browsers:

Full text

Close-range imaging, long-range vision

itut für
tions and
id Units
Cc Facial
nf. Proc.,
tween the
nent with
. Conf. on
nada, pp.
| Methods
ure and
3958, pp.
iding 2D-
on Laser,
cturing I,
juences of
16(2), pp.

Á. Detrekói ^", K. Fekete *, Z. Tóth *, O. Alhusain *" *, A. Juhász^, I. Stuber ©, Á. Rakusz
* BUTE, Department of Photogrammetry and Geoinformatics, Muegyetem rkp. 3. I. 19, H-1521 Budapest, Hungary (adetrekoi,
; kfekete, attjuhasz)@epito.bme.hu, alhusain@eik.bme.hu
MTA TKI, Geoinformatikai kutató csoport, Muegyetem rkp. 3. I. 19, H-1521 Budapest, Hungary
* Semmelweis Medical Sciences University
Commission V, WG V/3
KEY WORDS: Close Range Photogrammetry, Medical Imaging, X-ray.
In this paper, the determination of the measurements, shape and state of the human vascular system will be presented in detail. The
determination process is done through creating and plotting the topology of the human vascular system. The images used in this
project were, either collected by X-ray (róntgen) instrumentation or prepared by the correction of other imaging systems data. The
measurements done in the digital domain were mostly done by the use of general-purpose image processing software. The
measurements on the analogue images were done by the use of a monocomparator. For the displaying process, the split screen
method was chosen to display the normalized stereo pair photos. This procedure was chosen because it satisfies the medical
applications requirements for the need of displaying the plotting results onto various output instrumentation, and to assure wide
range applicability for the system.
During medical interventions to treat different illnesses, such as
brain surgery operations, it is extremely important if not life
saving, that the physician being able to know the dimensions
and structure of the human organs and their vascular network.
This exact knowledge about one human's organs can not be
generalized because it varies from one individual to another
making generalization in this regard associated with a good deal
of error possibility. During our investigation of similar imaging
methods like gamma cameras, computer tomography (CT),
Magnetic renaissance (MR), and classic X-ray (róntgen)
photography, we came to the conclusion that X-ray
photography was one of the most suitable methods. The
suitability of X-ray photography is due to its cost-effectiveness,
availability of X-ray instrumentation and expert personnel, and
the minimal intrusiveness associated with X-ray imaging.
In our study, which is a joint research effort of the Department
of Photogrammetry and Geoinformatics of Budapest University
of Technology and Economics (BUTE) and Semmelweis
Medical Sciences University (SMSU), the way of solving the
above-mentioned task was done by the use of analogue and
digital X-ray pictures. Object reconstruction which is
mentioned few times in this project depends on Direct Linear
Transformation (DLT) and was developed previously by our
team. The following sections will present this research project
in detail.
2.1 Data collection and pre-processing
During our work we had to solve the problem of placing the
control points, which in case of differently arranged pictures
ensure correct orientation possibilities. We have chosen a
solution well known from X-ray photogrammetry, that is, we
placed well-perceptible small metal balls on the plastic cassette

* Corresponding author(s).
holding the X-ray sensitive sheet at predetermined locations.
Pinning ensured the fixation of the horizontal coordinates that
are known from scale microscope measures.
X-ray photographs generally suffer from two kinds of
distortions inherently associated with X-ray imaging systems.
These distortions are contrast distributions degradations and
non-linearity distortions. Contrast distribution problems in
medical imagery can be corrected easily by traditional
sharpening filters such high-pass and high-boost filters
(Gonzales and Woods, 1993). Non-linearity distortions can be
corrected to an acceptable level by the use of 2-D interpolation
(Alhusain, 2000). In the 2-D interpolation algorithm, a cell built
on four neighbouring pixels is considered. The pixels are
represented by their coordinates and grey levels. Two 1-D
interpolations are carried out on the four pixels, two on each
side of the cell, resulting in two interpolated values; another 1-
D interpolation is carried out on the two previously interpolated
values giving in the final 2-D interpolation result (Castleman,
2.2 Measurements and object reconstruction
Measurements on the analogue photographs were done by the
use of PK-1 monocomparator. In the case of digital images, a
general purpose image processing system was used to measure
the coordinates of different points (pixels) in the photographs.
The accuracy of the measurements in relation to the digital
photograph area was in the sub pixel range. Another
measurement method, which substitutes the above mentioned
procedure, is the 3-D plotting in the spatial model created by
the normal stereo pair photographs. This last measurement
procedure requires that the related points in the two
photographs are identified and measured. In the measurement
process we used a program developed by the Geodetic and
Geophysical Research Institute of the Hungarian Academy of
Because of the special structure of X-ray imaging systems, the