May 15, 2008

The S-Curve Framework

By Weinstein, Ronald S

A challenge in business is to anticipate future technical advances as well as rates of diffusion of new technologies into the market place. Students of the history of science and technology are familiar with so-called S-curves of innovation. The S-curves show the relationship between performance and effort over time. The S- curves are classically sigmoidal in shape, thus the term S-curves (Figure 1). The S-curve framework can be useful for forecasting market trends as well as for understanding human perceptions of innovations. Arthur C. Clarke, the famous writer once noted, "One tends to be overoptimistic in the short run and underoptimistic in the long run. We can only extrapolate linearly and progress is always in an exponential curve." 1 Much has been published on the topic of innovation.2 The implementation of telepathology and, more recently, pathology digital imaging, an enabling technology for virtual slide telepathology, may be following an S-curve. It is too early to know for certain, but this would appear to be the case. The recent increase in interest in telepathology and pathology digital imaging, and the accelerating growth of the virtual microscopy industry, all of which are interrelated, suggests that telepathology is entering a new era in its level of use in laboratories.

It may be noteworthy that the development and implementation of digital radiology and teleradiology are following an S-curve for their implementation following a long interval between the initial work on developing the technologies in the laboratory and their eventual widespread deployment in clinical settings. Pathology could follow radiology's example and go entirely digital sometime in the future. It is possible that teleradiology and telepathology may end up following similar curves of innovation with pathology's time line trailing radiology's time line for going digital by a decade or two. On the other hand, somewhat different market forces drive radiology and pathology practices.

Telepathology was first used in the 1960s (so-called television microscopy) but did not begin to achieve sustained growth as a clinical imaging modality until the 1990s. For all practical purposes, events in the late 1980s involving the introduction of robotic telepathology, store-and-forward telepathology, and the development of hybrid telepathology systems, integrating robotic dynamic telepathology with store-and-forward telepathology, marked the initiation of the S-curve of innovation for virtual slide telepathology. Some current drivers that increase the use of virtual slide telepathology include (1) the growing base of installed virtual slide scanners, (2) incorporation of virtual slides into education programs, and (3) the emergence of new laboratory business models and services that leverage virtual slides as their enabling technology.

In Arizona, and in my previous department in Chicago, interest in telepathology was initiated by concerns regarding pathologist interobserver variability in rendering surgical pathology diagnoses. The first scientific paper using the word telepathology was presented at a College of American Pathologists conference entitled "Pathology Practice in a World of Changing Technology," January 16 through 18, 1987, held in Hawaii.3 This Futurescape conference we are attending here in Chicago marks the 20th anniversary of that earlier College of American Pathologists conference.

This conference provides me with an opportunity to discuss where we are in relation to the S-curve of innovation for telepathology and virtual slide microscopy. The introduction of virtual slide telepathology will have a significant impact on both telepathology and pathology digital imaging.

Historically, low-resolution video whole slides (virtual slides) were used in my laboratory in Chicago in 1986 for purposes of orienting histopathology slides and tracking slide viewing by telepathologists using a robotic telepathology system that I invented and patented.3 That invention and its initial testing were described at the 1987 College of American Pathologists Foundation conference in Hawaii.

Some aspects of virtual slide telepathology (also called whole slide imaging) were independently developed by my research group and by James Bacus, PhD, in the 1980s and early 1990s. He and I were collaborators at Rush Medical College in Chicago in the late 1970s and early 1980s, when I was pathology department chairman at Rush Medical College and he was a senior scientist at Rush. We each had our own laboratories but collaborated on his automated cytopathology project. We coauthored an abstract on automated cytopathology with Dr George Wilbanks. To produce my group's video whole slides, we used a low-resolution video camera mounted on a light box. To produce high-resolution digital virtual slides, Jim Bacus' group used a single optical axis light microscope and electronically imaged a glass slide 1 microscopic field at a time. The images were electronically stored. His solution was far more elegant, but processing times for the first Bacus virtual slides were many hours, whereas the production of our low-resolution video slide files was nearly instantaneous, as required for use for histopathologymicroscope field tracking in the Corabi robotic telepathology system developed and tested by my group.

Virtual slide scanners incorporating single optical axis optics as their digital imaging engine have become much faster since then. Additional decreases in processing times can be expected with the invention of new optical systems that dramatically increase the size of a field of view of a light microscope operating at relatively high magnifications, such as x 20 or x 40. How does such a system work?

Because this meeting is called "Futurescape of Pathology," I am going to discuss what our group at the University of Arizona is doing in optics research and development on what we regard as a "next generation" scanner. We are working on a virtual slide scanner based on novel very large field-of-view optics. The scanner incorporates a new technology developed in Arizona, the array microscope. For purposes of disclosure, I am a coinventor of the array microscope. My colleagues and I have been issued fundamental US patents for it.4,5 The technology is being commercialized by a spin-off company of the University of Arizona, DMetrix, Inc (Tucson, Ariz).

In the year 2000, I was approached by Michael R. Descour, PhD, an outstanding optical scientist at the Arizona College of Optical Sciences (then called the Arizona Optical Sciences Center) and Peter H. Bartels, PhD, a distinguished professor of optical sciences and pathology at our institution. Bartels had proposed that, through miniaturization, tiny microscopes might be produced and aggregated into an optical imaging "chip" (miniaturized "lenslet array ensemble"), which could serve as a digital imaging engine for a very rapid virtual slide scanner. This could theoretically process a virtual slide in under a minute. In 2000, that represented a 20- fold or greater improvement in virtual slide processing times.

Bartels realized that it would be possible to increase the field of view of a conventional light microscope from a millimeter (using a x 20 objective lens) to several centimeters using such a miniaturized microscope array. Because the field of view of his proposed optical device could be the width of a glass slide, it would be possible to digitally image a whole glass slide with a single pass of the digital imaging unit. Massive parallel processing of data would further reduce the processing time for a virtual slide.4,5

Fabrication of such an optical device would involve producing miniaturized optical benches, about 1 cm in height. These would, in turn, incorporate tiny lenses. Producing such a device would require the use of aspheric optics, which was regarded as quite challenging. However, the use of aspheric optical elements was critical for success because it would increase the field of view of individual lens from approximately 1 to 25 field of view per lens diameter for a spheric lens to 1 to 7 for an aspheric lens. This very large field of view, relative to the diameters of the tiny lenses in each lens system, would permit the packing of these tiny optical benches into compact geometric arrays.4,5

Another technical challenge came from the requirement to precisely align the small optical benches in an axis perpendicular to a glass slide to achieve uniform focus of histopathology sections. A unique design feature of the DMetrix virtual slide scanner imaging system came from research I participated in, 4 decades earlier, in the field of membrane biology. At that time, a colleague, N. Scott Mc- Nutt, and I were carrying out ultrastructure research on freeze-fractured gap junctions at the Massachusetts General Hospital in Boston. In 1967, when we initiated this research, I had a dual role as a 28-year-old independent National Institutes of Health-funded director of the Mixter Laboratory for Electron Microscopy, on the neurosurgical service at the Massachusetts General Hospital, and as a resident in the department of pathology, chaired by Dr Benjamin Castleman. Scott McNutt, who first-authored the key paper, was also a Massachusetts General Hospital pathology resident, a year behind me in the pathology residency training program.6 In retrospect, in the late 1960s, Scott McNutt and I were young men, absorbed in the major biology research issues of the day, who were fitting serious research into our busy schedules as Massachusetts General Hospital pathology residents. Because of our biologic research interests in membrane ultrastructure, Scott McNutt and I became familiar with planar arrays of functional units in biomembranes. In the case of biologic gap junctions, protein oligomers containing transmembrane channels mediate metabolic and electronic coupling of cells.6 Decades later, it turned out that the DMetrix engineers in Tucson, Ariz, were able to apply some of the same structural principles found in nature at gap junctions to the design of our novel array-based optical system. Our recent use of planar arrays of lenses at DMetrix provided a solution to the Z-axis alignment issue for the individual imaging units in the DMetrix array microscope (Figure 2). Fabrication of the DMetrix imaging device was greatly simplified by substituting 3 planar monolithic lenslet arrays (Figure 3) for the 80 to 100 perpendicularly oriented tightly packed miniature optical benches, as originally envisioned by optical scientists for the design of the miniature microscope array.4 This evolution in thinking about microscope optical design, and the progression from bundled miniaturized optical benches to fabrication of lenslet array ensembles, is illustrated in detail in our US patent for the array microscope.5

The DMetrix array microscope optics consists of a stack of three 80-element 10 x 8 arrays, constituting a lenslet array ensemble. The ensemble is approximately the same size as a small stack of US quarter coins (Figure 3). A CMOS (complementary metal oxide semiconductor) sensor (not shown) is attached to the top of the lenslet array ensemble. The device serves as the digital imaging engine of the DMetrix ultrarapid virtual slide scanner.4

In operation, the DMetrix DX-40 optical system moves along the long axis of a histopathology slide at about 3 mm/s and accrues about 240 000 digital images a second. These are assembled into a large digital image file by massive parallel processing. As soon as a glass slide is fully scanned, the large digital image file is posted on a server and is available on the Internet for examination from a distant location by a pathologist using a virtual slide viewer (browser).

Today, a DMetrix ultrarapid virtual slide scanner is routinely used for clinical services and teaching of medical students, residents, and fellows at the University of Arizona in Tucson. The successful implementation of ultrarapid processing has underlined the value of achieving high virtual slide throughputs in the processing of virtual slides for clinical services. Other companies have accelerated the development of their higher speed scanners as well. These developments have resulted in plans to expand virtual slide telepathology services and to explore a number of new applications of virtual slides. Although only a handful of laboratories had virtual slide scanners a decade ago, installations of these scanners now number in the hundreds and could be in the thousands in the foreseeable future.

Returning to our discussion of the S-curves, we might anticipate that plots of S-curves for virtual slide telepathology will show dramatic increases in numbers of virtual slides produced in pathology laboratories worldwide during the next few years. First of all, the number of scanners available for scanning has increased making the technology more readily accessible. Virtual slides also are being used for more applications. For example, although exact numbers are not available, it is likely that the majority of US medical students are using virtual slides, over theWeb if not at their own institutions. The American Board of Pathology uses virtual slides for its primary certification examination. A number of cancer centers are exploring the uses of virtual slides for pathology validation studies for clinical trials. The pharmaceutical industry appears to be embracing virtual slide technologies. New laboratory services are becoming available based on virtual slides. With the "US Labs business model," the technical component of specimen processing can be done at a reference laboratory, whereas case readouts are handled by a telepathologist working from the referring laboratory.

The use of virtual slides could also further accelerate if new clinical practice models are developed based, in part, on the technology. At the University of Arizona, we are piloting the implementation of a new bundled rapid breast care service (Figure 4). A breast abnormality may be detected by imaging or palpation. Digital mammography results are rendered either by a digital mammographer onsite or by an off-site teleradiologist. A core biopsy can be performed immediately, either by the radiologist or a surgeon. Tissue is rapidly processed into histopathology slides in a matter of hours. Virtual slides are processed at the laboratory and read out by a telepathologist, often at a distant site, over the Internet. A patient with a positive biopsy may meet immediately, using bidirectional video conferencing, with a teleoncologist for same-day treatment planning.7 At the University of Arizona, more than 300 women have received same-day breast biopsy results. Patient and service provider satisfaction is high.

At the Arizona Health Sciences Center in Tucson, which houses the University of Arizona College of Medicine, DMetrix-scanned virtual slides are being used for many different applications. They are used for cognitive psychology research aimed at identifying biomakers of expertise. 8 We use DMetrix virtual slides to teach medical students and residents on a daily basis. The University of Arizona has 2 primary teaching hospitals in Tucson. Virtual slides are used at quality assurance sign-out sessions enabling staff members located at both hospitals to participate in these consolidated multi- institutional surgical pathology conferences. Virtual slides are also used to show cases to staff subspecialty pathologists when they are out of town. Subspecialty pathologists can also examine surgical pathology cases using the secure Internet when they are in other countries attending international meetings or on sabbatical leaves, provided that they have access to approved facilities for remote viewing. Use of virtual slide telepathology has increased University of Arizona pathology staffing flexibility as well as access to subspecialty pathologists. We continue to have a high level of staff pathologist job satisfaction, fostered in part by a shared sense of being leaders in anatomic pathology innovation.

Extrapolating from the University of Arizona's recent experiences, and the rapid expansion of the base of installed virtual slide processors worldwide, I would anticipate that the production of virtual slides will increase in the years ahead. It remains to be seen if virtual slides will totally replace glass slides anytime in the foreseeable future. On the other hand, the advantages of digital imaging, including its potential uses for spectral imaging and quantitative microscopy, might further encourage the adaptation of digital pathology in many laboratory settings in the future.


1. Weiner J. Predictions: 30 Great Minds on the Future. New York, NY: Oxford University Press; 1999:39.

2. Christensen CM, Raynor ME. The Innovators Solution. Boston, Mass: Harvard Business School Press; 2003:31-71.

3. Weinstein RS, Bloom KJ, Rozek LS. Telepathology and the networking of pathology diagnostic services. Arch Pathol Lab Med. 1987;111:646-652.

4. Weinstein RS, Descour MR, Liang C, et al. Reinvention of light microscopy: array microscopy and ultrarapidly scanned virtual slides for diagnostic pathology and medical education. In: Virtual Microscopy and Virtual Slides in Teaching, Diagnosis and Research. Boca Raton, Fla: CRC Press; 2005:9-35.

5. Weinstein RS, Descour MR, Liang C, Bartels PR, Stack RV, inventors; The Arizona Board of Regents, on behalf of the University of Arizona, assignee. Miniaturized microscope array digital slide scanner. US patent 7,184,610 B2. February 27, 2007.

6. McNutt NS,Weinstein RS. The ultrastructure of the nexus: a correlative thinsection and freeze-cleave study. J Cell Biol. 1970;47:666-688.

7. Weinstein RS, Lopez AM, Barker GP, et al. The innovative bundling of teleradiology, telepathology and teleoncology services. IBM Syst J. 2007;46:69-84.

8. Krupinski EA, Tillack AA, Richter L, et al. Eye-movement study and human performance using telepathology virtual slides: implications for medical education and differences with experience. Hum Pathol. 2006;37:1543-1556.

9. Weinstein RS, Descour MR, Liang C, et al. An array microscope for ultrarapid virtual slide processing and telepathology: design, fabrication, and validation study. Hum Pathol. 2004;35:1303-1314.

Ronald S. Weinstein, MD, FCAP

Accepted for publication January 11, 2008.

From the Department of Pathology, University of Arizona College of Medicine, Tucson.

Dr Weinstein is cofounder of and owns stock in DMetrix, Inc. He holds 2 US patents related to the DMetrix scanner, the rights of which are assigned to the University of Arizona Board of Regents. Dr Weinstein also is cofounder and owns stock in UltraClinics, Inc. He has applied for a US patent on the ultraclinics process through the University of Arizona Board of Regents.

Presented at the College of American Pathologists Futurescape of Pathology Conference, Rosemont, Ill, June 9 and 10, 2007.

Reprints: Ronald S. Weinstein, MD, FCAP, Department of Pathology, University of Arizona College of Medicine, 1501 N Campbell Ave, Tucson, AZ 85724-5043 (e-mail: [email protected]).

Copyright College of American Pathologists May 2008

(c) 2008 Archives of Pathology & Laboratory Medicine. Provided by ProQuest Information and Learning. All rights Reserved.