Radiology imaging goes 3-D

Biomedical researchers at the National Institutes of Health are excited by the
prospect of digital radiology workstations that can analyze 3-D images in real time.


For the past year, the NIH Center for Information Technology has been developing
software to analyze 3-D echocardiograms (ECGs) on Silicon Graphics Inc. Onyx2
InfiniteReality workstations. The software makes analyzing ECGs more accurate, NIH
computer engineer Raisa Freidlin said.


Once captured, the 3-D images are available for later comparisons. “A few cycles
of the beating heart are stored on hard disk, and the physician can go back and
re-evaluate,” Freidlin said.


Using real-time 3-D images and parallel computing methods, NIH researchers “can
see more things of medical interest,” said Robert Martino, chief of the Computational
Bioscience and Engineering Laboratory in the Center for Information Technology in
Bethesda, Md.


Conventional ECG studies rely on composites of 2-D image slices. The new technology
lets researchers outline with a mouse in 2-D and propagate the results into 3-D to
represent the area accurately.


Physicians can better estimate the size of an infarcted area, for example.


If a transducer fails to capture a particular plane during a 2-D ultrasound procedure,
the plane is lost to medical evaluation. But with real-time 3-D visualization,
cardiologists can view any plane at any time, Freidlin said.


To develop the visualization algorithms, Freidlin uses an eight-processor SGI Onyx2
InfiniteReality rackmounted workstation with 4G of RAM and 64M of texture memory.


The InfiniteReality has a high-speed hardware connection between its processors and the
screen display.


“As we build more general applications, texture memory will let us display bigger
data sets,” Freidlin said.


She uses C, C++ and OpenGL languages to develop the medical algorithms. While the
work is still in its early stages, NIH cardiologist Julio Panza uses the tools on a
two-processor SGI Onyx2 InfiniteReality deskside workstation to evaluate the patients he
sees daily at the National Heart, Lung and Blood Institute.


“3-D reconstruction was a very cumbersome process,” Panza said, but real-time
3-D images generated by the newest echocardiographs require no reconstruction.


Radiologists prefer ultrasound studies because they are noninvasive and do not expose
patients to radiation.


But physicians are not completely satisfied with real-time 3-D images, Martino said.
They want to get into the fourth dimension—time-series images of beating hearts.


The 3-D ultrasound project is one of many applications for parallel computing in
biomedical research, Martino said.


“Through the years, we’ve applied parallel computing to structural biology,
bioinformatics” and other computation-intensive problems, he said. Now, medical
imaging “is really exploding.”  





inside gcn

  • artificial intelligence (vs148/Shutterstock.com)

    Government leans into machine learning

Reader Comments

Please post your comments here. Comments are moderated, so they may not appear immediately after submitting. We will not post comments that we consider abusive or off-topic.

Please type the letters/numbers you see above