"Visualization
in Medicine:
Three-Dimensional Magnetic
Resonance
Breast Imaging"
Tomasz
Macura, Mathematics and Computer Sciences, UMBC,
Advisor: Dr..
Chein-I Chang, Professor of Electrical Engineering
Remote Sensing Signal and Image Processing Laboratory (RSSIPL)
Department of Computer Science and Electrical Engineering
University of Maryland Baltimore County (UMBC)
1000 Hilltop Circle, Baltimore, MD 21250
The aim of this project
was to develop software for the three-dimensional visualization of the
female breast with cancer lesions based on data derived from the magnetic
resonance imaging modality. Breast cancer is the most frequently diagnosed
and the second most lethal cancer for American women. Currently, despite
intensive research, no cure or vaccine for breast cancer exists. The only
known recourse for reducing the mortality of breast cancer is early detection.
Although it is frequently described as a diagnostic exam, mammography
is primarily used for screening. Many benign lesions detected on mammograms
are indistinguishable from their cancerous counterparts by any known form
of noninvasive assessment. Once a lesion is detected, the radiologist
usually recommends that the patient undergo a biopsy. During biopsy, a
sample of the suspicious tissue is obtained and examined for pathologic
evidence of cancer. It would be beneficial if a computer could guide the
tissue sampling. With increased accuracy, smaller pieces of the breast
could be removed during biopsy. This would make the procedure less invasive
and traumatic.
Prof. Chein-I Chang and I worked on developing three-dimensional visualizations
from magnetic resonance (MR) breast scans. In its raw state, MR data is
a set of two dimensional parallel grayscale images called slices. In order
to create a three-dimensional model we converted the series of 60 MR slices
into a three-dimensional organ model. Next we created a software-based
visualization framework that allowed us to display the three dimensional
model onto a two dimensional monitor. This was done by defining an opacity
function. Our software allowed us to interact with the model (translate,
rotate, and zoom) and see the object from different perspectives. Our
program was developed in house by the author, using the C++ programming
language and VTK C++ Libraries (Visualization Toolkit, Kitware Corporation,
Clifton Park, New York, U.S.A.)
