In 1984, Dr. Soll and two colleagues (Dr. S. Kater and the late Dr. M. Solursh) in the Department of Biological Sciences at the University of Iowa successfully applied for a program project grant from the Institute of Child Health and Human Development at NIH to study the developmental biology of cell motility in three biological systems. A core facility was to be developed which would provide state-of-the-art microscopy, image processing and motion analysis software. At the time of the review, members of the NIH panel felt that the PC motion analysis software developed by Varnum and Soll was inadequate for the expanding needs of the three basic research projects, and recommended that an image processing system purported to perform automated motion analysis be purchased. After a day of testing, Dr. Soll, the Principal Investigator of the program, concluded that the system did not and could not automatically analyze the behavior of living cells. It was, therefore, decided by the Program investigators that the staff of the Core Facility attempt to design and implement its own system to fulfill the objectives of the Program. To that end Dr. Edward Voss, an expert programmer with a Ph.D. in theoretical mathematics, was hired. Within four months of the time Dr. Voss and Dr. Soll had begun setting up the Core Facility and had begun designing a motion analysis software system, a company announced the availability of the first bona fide automated motion analysis system, Expertvision. It included a videorecorder, a real-time contour digitizer and UNIX-based software in a SUN computer. High contrast videorecordings of cells were played into the digitizer, which identified pixels at high contrast cell boundaries. Boundary pixels were on and all other pixels off. The x,y coordinates of these pixels were stored in a UNIX data file and used to compute the position of the cell centroid (center of boundary) at time intervals as short as a thirtieth of a second (i.e., video rate). Dr. Soll tested and then purchased the first system sold by this company, but quickly discovered that although valuable for studying velocity and direction, it had a major limitation for studying animal cells. It literally threw away the perimeter information once the position of the cell centroid was computed. In all three basic research projects, animal cells represented the experimental organisms and shape changes were basic to understanding cellular locomotion. The company allowed the Iowa researchers to access their datafile formats, in order to develop a software program that would analyze the dynamic morphology of animal cells. The Iowa researchers first sped up the system. They next developed algorithms that measured dynamic contour changes.

This program, finished in 1986, represented the first computer-assisted dynamic morphology system, DMS (Figure 1A).

By 1988, Drs. Soll and Voss realized that DMS software had too many limitations. First, it relied upon automated digitization of the cell perimeter by the contour digitizer. This required a high contrast cell image and, therefore, excluded the use of differential interference contrast (DIC) microscopy. It also excluded the analysis of flat cells, like fibroblasts, which could not be sufficiently contrasted for edge detection. Expertvision offered no alternative manual digitization option for low contrast cells. Second, the digitizer produced an artifact when interpreting the edges of narrow objects. Third, DMS software had to access large UNIX files in order to function, which slowed the system significantly. Fourth, the program had few of the newly developed advantages of the Macintosh computer operating system available in 1988. Fifth, it was becoming clear that 2D analyses were insufficient in some situations, and that a 3D system would be necessary to answer a variety of motility questions. This last problem was becoming especially acute in research projects on the cell motility and chemotaxis of human white blood cells and D. discoideum amoebae, in which it was clear that portions of cells changed focal planes (i.e., went in and out of focus) as the cells crawled and responded to chemoattractants and other extracellular signals. Therefore, the decision was made in 1988 to develop 1) a Macintosh-based 2D system with both automatic and manual modes of digitization, 2) an artifact-free contour digitizer, and 3) a 3D dynamic image analysis system. Dr. Soll, therefore, obtained funds from the State of Iowa for technology development. A prototype of an artifact-free real time contour digitizer, the ACD-101, was completed in 1993, but was immediately deemed obsolete because of the parallel evolution of less expensive frame-grabbing technologies and increased computer speeds, allowing superior software-based, automatic outlining of cells. The first 2D dynamic image analysis system based on a Macintosh computer (2D-DIAS) (Figure 1B) was completed in 1993, which used a frame-grabber board both for automatic and manual digitization.

In 1993, the first 3D-dynamic image analysis system (3D-DIAS) prototype was also completed (Figure 3). It utilized both a Tektronix 3D stereo workstation with a dedicated Tektronix computer for 3D imaging, and a SUN-4 computer for computing 3D motility and shape parameters. This original system (3D-DIAS I) was effective in generating dynamic 3D reconstructions (i.e., 3D movies) of crawling cells, but required laborious manual tracing of every optical section in the reconstruction process, thus limiting the number of sections per reconstruction. The original 3D-DIAS I system also lacked an effective computer-driven stepper motor with a synchronizing character generator for accurate optical sectioning. Sectioning was performed manually with a precalibrated focus knob and a metronome. Synchronization and height calibration were accomplished by using single frame stepping on the VCR with calibrated bead markers in the preparations.

By 1993, it was realized that 2D-DIAS needed significant updating and refinement, and that 3D-DIAS needed to be transferred to a new workstation since the Tektronix workstation and SUN computer had become obsolete. Servicing the Tektronix equipment had become increasingly difficult and expensive, and down-time had by then exceeded up-time. Dr. Solls staff, therefore, began to transfer 3D-DIAS to a Macintosh-based system. In addition, they began to experiment with new methods to increase the edge detection capabilities of both 2D-DIAS and 3D-DIAS software, and to build a computer-regulated stepper motor for optically sectioning cells that included a character generator that labeled each frame for height, time and direction of scan. The original 3D-DIAS software program was composed of an estimated 300,000 lines of C-code in 1993, so the transfer to a Macintosh-based system was a significant undertaking. However, in recreating the system, several major advances and inventions were added that included rapid, accurate and totally automated edge detection based upon a pixel complexity measurement developed by Dr. Voss, direct image reconstructions, which saved all pixel information within the 3D cell interior, and a computer-regulated stepper motor, the MicroStepZ3D, for optical sectioning.

In 1997, Dr. Soll, successfully applied for a grant from the W.M. Keck Foundation for three years, which provided funds to set-up the following Facility components: 1) an Applications Laboratory containing six computer workstations with 2D-DIAS and 3D-DIAS software for users to analyze data; 2) an Experimental Laboratory containing two inverted microscopes with DIC optics and one near real time laser scanning confocal microscope, customized stepper motors at each microscope station for optical sectioning, videorecorders, 2D-DIAS and 3D-DIAS software for direct digitization, and customized environmental and perfusion microchambers; and 3) a Development Laboratory with two work-stations for software development and an engineering station for the construction of stepper motors and cell chambers, and for the maintenance of computers and equipment.

With the funds received from the W.M. Keck Foundation, the capabilities of dynamic 3D reconstruction and analysis have been significantly expanded, the first high speed reconstruction system (up to 500 frames per second) has been initiated, the first dynamic 3D reconstruction and analysis software has been written for confocal images, advanced stepper motors have been developed for rapid optical sectioning, and the first 2D-DIAS program has been customized for echocardiography (2D-DEAS). In addition, the Facility has supported visits in the past year by thirteen scientists from outside the University of Iowa for periods of up to four months and several projects by researchers from within the University.