| |
News
& Ideas
COMPUTER TECHNOLOGY:
The sound of lung disease
Michael Savic, professor of electrical, computer,
and systems engineering, and his graduate
student, Thrasos Axiotis, are developing computer technology that
will diagnose lung disease using signal processing.
Working with a professionally made database of
sounds of about 50 lung diseases, Savic and Axiotis have programmed
a computer to use features of these sounds to identify lung diseases,
such as pneumonia, asthma, and bronchitis.
| |
Working with a professionally made database
of sounds of about 50 lung diseases, Savic and Axiotis have
programmed a computer to use features of these sounds to identify
lung diseases, such as pneumonia, asthma, and bronchitis.
|
Lung sounds from the database are transformed
into electronic signals and brought into a computer, which then
analyzes the signal and determines ifthe lungs are healthy or
not.
“If you want to find someone with blue eyes
and a big nose, you look for those features, not for toes or hands
or hair,” Savic says. “Pulling features of particular
lung diseases works in the same way.”
If the lungs are not healthy, a graphical display
on the computer screen indicates the nature of the disease with
a specific color, called a cluster. There are presently a few
clusters that overlap, indicating that those diseases share some
of the same features, Axiotis explained. These will require more
refined signal processing to separate, he said.
The researchers plan to test their system on real
patients in the near future, and anticipate slight obstacles.
“The data we currently have is clear and
nice,” Savic said, “but in the exam room we don’t
have as much control over noise interference. Noise from an air
conditioner or even movement from the patient could distort the
signals, giving us an inaccurate reading.” But with a bit
of refining, the system will work, he says.
CONTACT: Theresa Bourgeois, (518) 276-2840,
bourgt@rpi.edu
|
