I worked on X-ray computed tomography systems for about 15 years, starting right out of college (my Master’s thesis was on flat-panel three-dimensional computed tomography). Over that time, I worked with a fantastic group of people to develop all kinds of custom systems. They were used in government installations, medical device manufacturers, and we even made medical systems for placement in dentists’ offices for full 3D imaging of patient’s heads, something that is common now but was brand-new at the time.
My favorite system was one of the last we developed, and we named it “Sprite”. It was a smallish cabinet unit with fairly ordinary specifications for an industrial X-ray CAT scanner at the time. But what made it amazing was that it was designed from the start to be “as easy to use as a copier machine”. The idea was that someone with five minutes of general guidance could turn it on and make a full CAT scan of any object that would fit into its sample holder. They would know how long the scan would take, and it would fully self-calibrate. Then the system would process the final data into a three-dimensional view they could easily slice through or spin around.
Sprite used an integrated PC and used the GPU for data processing (again, a new idea at the time). It could do in 20 minutes what used to take days. What we found was that when we made the system simpler to use, we really made it better. We could routinely make better imagery than a big expensive CAT scanner run by an expert with years of experience. Developing Sprite required rethinking everything. We redesigned how the scatter prevention systems worked. We wrote a completely new touch screen interface (on Windows XP!). We used an RFID system to automatically detect calibration objects and which size sample holder was inserted. And we designed a completely new way to fabricate the shielding enclosure, which allowed for a smaller cabinet with better ergonomics.
