University of Washington

Building A Better Pre-Clinical PET Imaging System

---

In the Department of Radiology at the University of Washington Medical Center, Professor Tom Lewellen and his team are designing Mac-based imaging systems that give investigators a closer look at metabolic functions in a mouse that weighs about an ounce.

“Biological investigators need higher resolution images,” says Lewellen, “partly because the regions of interest are so minute, and partly because they have to scale up what they see in mice to humans that may weigh 2500 times as much. So we’re using the Mac to design, build and run positron emission tomography (PET) scanners that push the resolution envelope.”

Lewellen, Director of the Nuclear Medicine Division’s Physics Group, is a longtime Mac user. The Group’s five faculty members, all of whom have NIH research grants, do their primary work on MacBook and MacBook Pro systems, as do the Group’s four post-doctoral fellows and three grad students, and many Radiology Department faculty members.

Building Better Scanners with the Mac

When the Physics Group began building their ambitious MiCES (Micro Crystal Element detector Scanner) five years ago, they started from scratch, using the Mac to design hardware and write code.

One obvious challenge was data communication. The design specified 72 photomultiplier detector modules communicating independently with a host computer that would generate DICOM images. The Physics Group looked at point-to-point and store-and-forward techniques. Then they looked at FireWire.

“Because of the way the Mac has implemented FireWire, we could write all our software in the user space,” says Lewellen. “That is incredibly important for a small group like ours, because writing software in the user space is much easier than kernel driver development.

“FireWire has been stable between upgrades. We haven’t had to modify our basic code for FireWire transactions from OS X 10.1 all the way up through 10.5. That has saved us a huge number of man-hours.” MiCES uses an 8-core Mac Pro running Xserve for data acquisition and initial image reconstruction. It has 700-micron resolution, compared to the 1.3mm resolution of the university’s newest commercial small-animal PET scanner — nearly a factor of 2 improvement.

Mac OS X with UNIX: Best of Both Worlds

The Physics Group currently uses 15 MacBook Pro and six iMac systems to design circuit boards, write code, and simulate scanner systems under development. They run simulations on an eight-core Apple Xserve cluster with seven terabytes of RAID. An 8-core Mac Pro with two Apple Cinema Displays is used exclusively for viewing scanner images with OsiriX, an open-source DICOM viewer.

“The big thing is to have the joys of the Mac GUI and still be able to run UNIX code,” says Lewellen. “We didn’t have to rewrite lab tools written in UNIX — we just recompiled them under OS X.”

The Physics Group also does its field-programmable gate array (FPGA) and embedded processor code development on the Mac, running Windows XP-based tools in VMware Fusion or Parallels. Schematics and board layouts are created with MacCAD, a native Mac application.

The Agile MacBook Pro Does It All

Many Radiology Department clinicians, including the chairman, use the Mac as a primary computer. “I’m a researcher and a clinician,” says Department Vice Chairman Dr. Satoshi Minoshima. “I do everything — clinical work, education, research, administrative work — on my 15-inch MacBook Pro. If I want large-screen viewing, I plug my laptop into a 23-inch Apple Cinema Display.”

Using a university VPN, Minoshima does medical charting, reporting, case reviews, and all other job-related work wherever he happens to be, running Remote Desktop Connection or Windows XP in VMware Fusion on his MacBook Pro. He can access the Medical Center’s PACS and download images.

“The reason I can use the MacBook Pro for both clinical work and research is the UNIX background of Mac OS X,” says Minoshima. “I do research programming in the UNIX environment. I cross-compile everything on the MacBook Pro, whether it’s for Mac OS X, Windows, or Linux. I can be programming and cross-compiling while I travel internationally, in addition to checking my clinical work and email.”

Minoshima uses his MacBook Pro to write image analysis software as part of his NIH-funded research in Alzheimer’s disease. He provides his software to research communities worldwide and has licensed it for use on medical imaging systems.

“The discovery of abnormal brain tissue in the posterior cingulate cortex in 1995,” says Minoshima, “was made possible by image processing technology developed on the Macintosh — just one of the ways Apple computers have contributed to better medical care.”

Ahead: More and Better Scanners

The Physics Group is currently at work on two more Mac-based MiCES scanning systems, including a small-animal PET scanner that will be inserted into a magnetic resonance imaging (MRI) system to provide co-registered anatomic and metabolic images.

“The payoff of our work with the Mac,” says Lewellen, “is helping our clinicians to get a better understanding of disease processes. We can follow the progress of diseases or therapies in a single group of animals rather than sacrificing them and taking tissue samples — a major efficiency.

“We’re often asked why we like the Mac better. The answer is that it’s easier to use. And when I go to the Worldwide Developers Conference and talk to Apple engineers I’m always impressed with the fact that these guys are really clever, and are given the freedom to be clever. We get caught up in that spirit and say, we can do something like that too.”