...
There are two major goals of this history: to document the events and conditions that led to the creation of one of the first grayscale image processors, and to describe the highly effective complementary collaboration that allowed this project to flourish. Occasionally, references will be made to other later advances indirectly related to the RTPP work that would not have happened without the RTPP. Where possible, we have linked to open access journal PDFs, and have included PDFs of the key technical reports describing the RTPP on this Web site.
| Anchor | ||||
|---|---|---|---|---|
|
The RTPP project was conceived and initiated by Dr. Lewis "Lew" Lipkin, M.D., head of the Image Processing Unit, later the Image Processing Section (IPS), in the the National Cancer Institute (NCI). The intellectual concept behind computer-controlled microscopy started in 1962 when Lew was an assistant professor of neuropathology at Downstate Medical Center in New York. Professor Patrick Fitzgerald, Chairman of the Pathology Department at Downstate, was studying pancreatic cell growth. Dr. Vinichaichol, who was doing visual grain counts on thin pancreatic sections, was finding mixed results. The problem was statistical. Dr. Lipkin was asked to design a proper sampling technique. Grain counting was a method used to measure cell metabolism before the days of antibody techniques applied to living cells and fluorescent techniques that came about during our time in NIH. Lew, who happened to know something about statistics, was asked by Dr. Fitzgerald to find out what was wrong with his statistics. After some thought, Lew realized that Dr. Vinichaichol was staying in one area of the slide and he had no way of knowing when he was recounting the same cells. Lew didn't want to continue looking at biological material that he couldn't explore without using some form of quantification.
...
The first thing he wanted to be able to do was move a slide via a computer-controlled microscope stage. Initially, he was going to do it with analog feedback. He talked to Wes Clark (who had helped build the LINC computer with Charlie Molner and others). Wes convinced Lew that he really wanted a digital stage - not an analog one - so that is what Lew developed: a series of stepping-motor-controlled stages that improved with each generation. The original design connected the stage with rubber bands, which was then greatly improved with direct stepping-motor drives. Lew had also been working with Russell Kirsch and Bill Watt from the National Bureau of Standards (NBS, now the National Institute of Standards and Technology or NIST 
). This early work involved describing biological images using computer picture grammars [1] that attempted to bring artificial intelligence and algorithmic methods to the description of biological images.
| Anchor | ||||
|---|---|---|---|---|
|
In 1968, I (Peter Lemkin) joined Lew's group to work on programming the LINC-8 along with Howard Shapiro of the PRB, and Russell Kirsch, Don Orser, and Phil Stein from the NBS who had been involved in the project. The first lab was in rental space in the Auburn Building across from the Bethesda Chevy Chase Rescue Squad where we would hear the fire trucks when they went out on a call. The group moved to the brand new Building 36 on the NIH Bethesda campus around 1970, which was a much better environment. (Building 36 was demolished in 2006.) The LINC-8 controlled a stepping-stage and a galvanometer scanner with a photomultiplier detector on a Leitz microscope, which was an early step in automated cytology [7]. It was very slow, but did offer high-quality 8-bit data. The problem was analysis power - in terms of scanning speed, CPU speed, image memory, analysis software, and analysis memory. It became clear that we did not have the hardware resources required to do complex image processing on the types of data we were determined to analyze. However, I learned to write hardware control software on the LINC-8 as it was truly a dedicated laboratory instrument computer ideal for connecting to laboratory equipment. This experience set the stage for the next generation of computer-controlled microscopes we tackled.
...
We were able to easily control the microscope and process the data using a small amount of PDP8e memory because of the computer language used. DEC's Fortran-II software language compiler running under their OS/8 operating system for the PDP8e allowed the insertion of assembly language that could reference I/O instructions (called IOPs) directly and could also directly reference Fortran variables. This made programming our new hardware relatively easy to do.
| Anchor | ||||
|---|---|---|---|---|
|
The plan was to have the NCI replicate these grain counter systems in three or four grantee laboratories. We had put out bids for the replication of the system. But, as with many technological break-throughs, the system worked, but better, less-expensive methods using new antibody and flow cytometry methods were becoming available. So autoradiography was replaced by other systems for measuring and quantifying specific cell types where tracking individual silver grains was not required. The additional grain counters were never built.
...
Overview
Content Tools
ThemeBuilder