The University of Wisconsin X-ray Microbeam Facility is a facility
for the acquisition of data concerning the dynamics of
speech.
It resides at the
Waisman Center
on the
University of Wisconsin Madison campus,
University of Wisconsin Madison
campus and was developed with
funds from the National Institutes of Health.
The facility is currently funded by
the grant:
``Articulatory Kinematics in Neurogenic Speech Disorders'' (R01 DC-03723),
funding source: NIH/NIDCD.
First use of the Microbeam system for data collection from human subjects occurred early during the 1987 calendar year. Between that time and the present, approximately 300 experimental sessions have been conducted on the system, involving roughly 200 different speakers and more than 40 research teams, yielding an aggregate data-set spanning more than 3200 tracking minutes.
The goal of the X-ray Microbeam is to acquire movement data
from human articulators during speech production simultaneous
with the acquisition of other physiological data over a relatively
long experiment session while minimizing exposure of the subject
to x-rays.
To achieve this, the Microbeam uses a narrow high-energy x-ray
beam to track gold pellets attached to articulators while
synchronously acquiring the physiological data.
One potential risk associated with X-ray Microbeam (XRMB) experimental procedures relates to exposure to ionizing radiation. In a typical database experiment, involving total aggregate tracking time of about 10 minutes, and, maximum exposure time per tissue volume of approximately 1.6 milliseconds per tracking second, we estimate that total peak effective dose equivalent will be on the order of 50mR.
Radiation exposures are relatively low with the XRMB system, for several reasons. The x-ray beam itself has a relatively small size, and is allowed to expose any given area of the oral cavity for only a brief time. In the worst case, the beam is on a given tissue area for 1.6 millisecond per tracking second. Electron beam current is minimized, producing an x-ray beam that contains only a sufficient photon density to image the pellets themselves. At the same time, beam energy is high enough to minimize soft tissue absorption. In general, beam power is optimized to provide only those photon densities necessary for reliable pellet tracking. Radiosensitive areas (e.g., the eyes and thyroid) can be avoided entirely, by disallowing x-ray beam deflection to those areas.
The best current estimate of the effective dose equivalent for 10
minutes of XRMB data collection is about 48 mR. For comparison, a single
dental whole-mouth exam yields an effective dose equivalent of about 10-30
mR (Laws, 1977), while a chest diagnostic exam results in an effective dose
equivalent of about 5-10 mR (NCRP Report #93, 1977). These comments are not
meant to imply that no risk is involved with XRMB procedures. Rather, they
indicate that the experimental XRMB radiation exposure risks relative to
other diagnostic flood-field techniques are comparable in magnitude and
quite low.
The Microbeam X-ray generator consists of a 450 KeV, 5 mA power supply, source electron gun and accelerating column, beam-line components for electron beam focusing and deflection, a thin ( 500 micron ) water-cooled tungstun target for photon generation, x-ray pinhole, and NaI detector. The entire machine is encased in 2-4 inch lead plates, making its weight approximately 15 tons.
It was built by the
U.W. Physical Sciences Laboratory .
The gold pellets (2-3mm diameter) are attached at
desired points
using an FDA approved dental adhesive (Ketac).
Pellets are in turn attached to threads for pellet retrieval
in case of aspiration.
Instead of irradiating the entire head during pellet tracking, the Microbeam generates a small rastered scan of the field where each pellet is expected to be. From each raster, the position of the pellet is determined and a prediction made as to where it will be at the next sample. The series of x-y positions for a pellet during a trial constitutes that pellet's trajectory.
A
schematic
shows details of the process.
After acquisition, acoustic and movement data can be plotted as a time-series. Movement data can also be plotted in two dimensions for a part of an utterance.
Several
data analysis tools
are available from Professor Paul Milenkovic.
A comparison of Japanese and American speakers is available on
Dr. Michiko Hashi's home page.
Gratuitous photo.
Principle Investigator: John Westbury
Laboratory Manager: Carl Johnson
Electrical Engineer: Rick Konopacki
Programmer: Jim Myers
Go to the
Waisman Center
on the U.W. campus.
The Microbeam is located in room 290.
You can park out front for fifteen minutes for free.
