The ACME Lab
(directed by Scott
Hauck) in the Electrical
Engineering department at the University of Washington and the Radiology
Division of Nuclear Medicine are collaborating to develop the
front-end electronics for a small animal positron emissions tomography
(PET) scanner. A PET scanner is a medical imaging technique that
produces three dimensional images or shows functional processes in a
subject. PET scanners
produce images by detecting radiation that is injected into the subject.
The short lived isotope is attached to a
metabolically active molecule so that it 
concentrates
in selective
tissues such as the heart, brain, tumors, etc. The radiation is
emitted in two anti-parallel photons that are detected by the scanner
ring . The detectors consist of
a scintillator that turns the emitted photon into visible light that is
then picked up by photomultiplier tubes (PMTs). The PMTs output
an analog signal to the front-end electronics. When two events
are detected by sensors on the opposite side of the scanners and
detected within a certain timeframe, the events are said to have
coincidence. The coincidental events are
digitized and sent to a host computer where the image is
constructed. (For more detailed
information on PET scanners check out this Wikipedia
page)
FPGAs and
PET Scanners
We are involved to refine and increase the functionality of the
front-end electronics. The current scanner (figure 3) has 18
detector modules in a ring. Each detector module has its own set
of electronics that digitize the signal, collect signal characteristics
(timing, strength, detector number, etc.) and creates a packet to be
sent to the host computer over a Firewire bus (IEEE 1394A). There
is also a single coincidence board that listens
to all 18
boards and
determines if events are in coincidence. Currently, the
electronics consist of a collection of discrete components such as
microcontrollers, ASICs, summing boards and FPGAs that we are going to
port to a single FPGA for the next generation scanner. We also
plan to increase the resolution of the coincidence timing so that
the scanner can determine where the event originated between the two
detectors based on different event arrival times (called time of
flight). Another challenge for the new system will be the
processing of up to 400 output channels from the ADCs to a single
FPGA. Finally, we want to be able to put the PET scanner inside a
MRI machine to facilitate simultaneous scans.
Pulse Processing in FPGA
As
the scanner is updated with newer generation FPGAs, more work will be
assigned to the FPGA such as pulse timing. Also, newer
capabilities will be added, such pulse pileup correction. Pulse
timing is the process of finding the starting point of a pulse such as
the one shown in Figure 4. Historically, pulse timing has been
performed by an analog circuit, but we are developing an algorithm to
perform pulse timing in
the FPGA. Of course, this means that pulse processing will
have used as the sampling rate of 14ns is
much
more coarse than the
desired timing resolution of about 1ns. The details of our
algorithm and results can be found in the two papers below.
Another pulse processing application that can be done in the FPGA is
pulse pileup correction. Pulse pileup occurs when two pulses
overlap because a second event occurs before the first event has
decayed back to zero. In order to use these two pulses, they have
to be separated. This work is in the preliminary stages.
M. D.
Haselman, S. Hauck, T.K. Lewellen, R.S. Miyaoka, "Simulation
of Algorithms for
Pulse Timing in FPGAs", IEEE Nuclear Science Symposium and
Medical Imaging Conference, 2007. (Better paper if you know PET)
M. Haselman, R. Miyaoka, T.
K. Lewellen, S. Hauck, "FPGA-Based
Data Acquisition System for a Positron Emission Tomography (PET)
Scanner", ACM/SIGDA Symposium on Field-Programmable Gate Arrays,
2008. (Better paper if you don't know PET well)
*This work is sponsored by Zecotek Medical Systems, NIH and
Altera.
*The principle investigator
is Thomas K. Lewellen, Ph. D., Director, Physics and
Instrumentation Development
This site is © Copyright
Michael Haselman 2004-2005, All Rights Reserved
Steve's free web
templates
