Implementation of Virtual Microscope Slides in the Annual Pathobiology of Cancer Workshop Laboratory

Transcription

1 Implementation of Virtual Microscope Slides in the Annual Pathobiology of Cancer Workshop Laboratory FRED R. DEE, MD, JOHN M. LEHMAN, PHD, DAN CONSOER, BA, TIMOTHY LEAVEN, MA, AND MICHAEL B. COHEN, MD Virtual slides are digital facsimiles of glass microscope slides that, when viewed with a pan and zoom viewer, can emulate viewing a glass slide with a traditional microscope. Based on successful implementation of virtual slides in medical student histology and pathology courses at the University of Iowa, we developed a plan to evaluate the use of virtual slides in the American Association for Cancer Research s annual Pathobiology of Cancer Workshop. In this Workshop, nonphysician predoctoral students and postdoctoral fellows working in cancer research explore the morphological, clinical, and molecular aspects of human cancer. Over the course of a week, students examine approximately 100 glass slides in microscope laboratories, facilitated by senior cancer investigators. The goal of the present study was to evaluate virtual slides as a teaching modality in Virtual slides are digital facsimiles of the visual content of glass microscope slides, acquired at a high magnification. Using a web-based zoom viewer, virtual slides can be magnified from to 40 and moved in a x y axis on the computer screen, thus emulating the examination of a glass slide with a traditional microscope. 1-6 Virtual slides are acquired by assembling digital images of adjacent fields in histological sections or fluid smears into a giant image montage using a computercontrolled scanning stage. Although this technology has existed for more than a decade, this process (also called whole-slide digitization) has been commercially available for standard desktop computing only since 1998, when sufficient memory and processing speed became available. Because virtual slides can capture the important visual information of a glass slide, they can be used in all of the educational venues in which glass slides and digital photomicrographs are used, without losing the ability inherent in traditional glass slide microscopy to explore the entire microscopic section, identify structures and lesions, and demonstrate their relationships and variations. From the Department of Pathology, Carver College of Medicine, University of Iowa, Iowa City, IA, and the Department of Pathology, Brody School of Medicine, East Carolina University, Greenville, NC. Accepted for publication March 13, Supported by a National Library of Medicine Information Systems Grant (G08 LM ) and a grant from the National Cancer Institute (5R25CA ). Address correspondence and reprint requests to Fred R. Dee, MD, 100 Medical Laboratories, Room 1198, University of Iowa, Iowa City, IA Elsevier Inc. All rights reserved /03/ $30.00/0 doi: /s (03) these laboratories, not as a replacement for traditional microscopy, but rather in terms of their utility in facilitating student learning as they examine glass slides with a traditional microscope. Evaluation by questionnaire indicated that virtual slides enhanced students ability to grasp morphological features better than the traditional photomicrographs. The results of this implementation suggest that virtual slide technology may be successfully extended to other educational venues where traditional microscopy and photomicrographs are currently used. HUM PATHOL 34: Elsevier Inc. All rights reserved. Key words: virtual slides, virtual microscopy, education, histopathology, cancer. Abbreviation: AACR, American Association for Cancer Research. In 1997, realizing the important contributions that virtual slides might make to education in microscopic anatomy and pathology, we implemented a plan, which was subsequently supported by a National Library of Medicine Information Systems Grant, to digitize all of the slides in the histology and pathology courses at the University of Iowa. These virtual slides were then successfully incorporated into the histology and pathology courses in 2000 and 2001, where they are used to augment traditional glass slide microscopy. 3,7,8 Based on this positive experience, we decided to evaluate virtual slides as a teaching modality in the microscope laboratory at the American Association for Cancer Research s (AACR) annual Pathobiology of Cancer Workshop, in which 3 of the authors serve as course faculty. The goal was not to replace the use of glass slides and microscopes, but rather to augment the ability of course faculty to facilitate learning as students examine cancer histopathology slides with traditional microscopy. The AACR Pathobiology of Cancer Workshop is a 1-week course that had its inception in Supported by the National Cancer Institute (NCI) since 1982, the Workshop provides scientists working in cancer research with an intensive experience in the morphological, clinical, and molecular aspects of human cancer. The course attendees are primarily nonphysician predoctoral students and postdoctoral fellows, and 5% are senior scientists. The Workshop consists of 7 half-day gross and microscopic pathology laboratories, interspersed with lectures on current concepts of cancer, including papilloma viruses, transgenic mouse models, cancer progression, metastasis, stem cell origins of cancer, molecular diagnosis, cellular and molecular therapy, and clinical trials. Six groups of 18 students each attend a laboratory session devoted to normal histology and in- 430

2 VIRTUAL MICROSCOPE SLIDES (Dee et al) FIGURE 1. A Keystone laboratory in progress. Dual projectors in this laboratory allowed projection of virtual slides and PowerPoint slides on the screen at the right, and gross images, diagrams or supplemental photomicrographs on the screen at the left. Students are seated at tables with microscopes and sets of glass slides. (This figure has been electronically enhanced). troduction to cancer, and then rotate through six 4-hour laboratory sessions covering cancers of the major organ systems. These 6 laboratory sessions are organized as (1) hematopoietic tissues and AIDS; (2) gastrointestinal tract; (3) childhood tumors and brain; (4) breast, uterus, cervix, and ovary; (5) male genitourinary tract; and (6) lung and skin. Each laboratory is staffed by at least 2 faculty members. In 2002, the Workshop faculty comprised 15 senior cancer investigators from 7 different US medical centers. Most of the 15 have taught in the Workshop for more than 3 years, and 1 faculty member has participated since the course s inception in 1978, lending stability to the Workshop. Learning aids in the laboratories include singleheaded binocular microscopes (1 per student), microscopic glass slide sets (approximately 14 slides per laboratory session), one double-headed microscope, gross specimens, overhead transparencies, 2 2 carousel slides, Microsoft PowerPoint (Microsoft Inc, Redmond, WA) presentations, and, more recently, virtual slides, which use the same computer projection equipment as PowerPoint. A typical laboratory is illustrated in Figure 1. In the laboratories, using a traditional microscope and glass slides, students identify the characteristic findings of cancer including dysplasia, in situ carcinoma, invasion, metastasis, abnormal mitotic figures, hyperchromatism, anaplasia, and differentiation, while comparing and contrasting normal, hyperplastic and neoplastic tissues. Students also identify the characteristics of individual cancer types for each organ system. Studying the histology and gross pathology of specimens concurrent with information related to the pathogenesis and pathobiology and clinical features helps the students develop a global concept of cancer. Use of the traditional microscope is an integral part of the course. This is not only because many of the trainees will be evaluating histologic slides in their own research laboratories, but also because the course faculty believe that the active learning associated with examining real glass slides is critical to developing an understanding of the pathobiology of cancer. The typical laboratory is organized as follows. Faculty members give a brief introduction to the concepts and content of the laboratory session, then introduce the clinical and gross features of the cancer, followed by the microscopic characteristics that the students are expected to find on their own glass slide. The instructors then circulate to assist students as they examine the slides and to discuss concepts of cancer exemplified by the laboratory materials. Many students in the Workshop have had a minimal amount of microscopic morphology. Thus the faculty s ability to demonstrate important features on the projector screen at the front of the laboratory is critical to the learning process. Before 2001, teaching aids for microscopy morphology consisted of 2 2 slides projected by a carousel projector or PowerPoint computer projector. Videomicroscopy was occasionally used, because it allowed the faculty to project the actual side the students were examining; however, this modality proved too cumbersome for routine use. The doubleheaded microscope is used sparingly for individual students when a specific issue requires direct interaction between a student and a laboratory director. To test the feasibility of using virtual slides as a teaching aid in the Workshop laboratory, we implemented virtual slides and evaluated them in 1 laboratory unit (male genitourinary tract) in the July 2001 laboratory. Based on the positive 2001 formative evaluation from this laboratory (described later in this article), we created a website for the course, and fully implemented virtual slides in all of the Workshop laboratories in July MATERIALS AND METHODS The method outlined here describes the technology used in to develop virtual slides for the AACR course. It is similar to that used for other Iowa virtual slide educational programs (see virtualslidebox/). 3,7,8 We anticipate that the technology used for future iterations of the AACR course will change as virtual slide technology evolves. Acquisition Visual slides were created with a system using Virtual Slice (MicroBrightField, Williston, VT) and ScanScope (Aperio Technologies, Vista, CA) technologies. The system included a panning motor, autofocus high-resolution microscope lens, optical capture device, computer, and software that automatically coordinates the activities of the motor and video capture. The equivalent of 1200 contiguous 40 microscope fields were automatically digitized, merged into 1 seamless image montage, and outputted as a single gigabytesized uncompressed.tif file of approximately 20,000 20,

3 HUMAN PATHOLOGY Volume 34, No. 5 (May 2003) 432

4 VIRTUAL MICROSCOPE SLIDES (Dee et al) pixels. Each file captured approximately 0.9 cm 2 of the surface of the glass slide. Actual digitization time ranged from about 15 to 90 minutes, depending on the equipment and configuration used. Although virtual slides prepared for the 2002 Workshop were limited in size and speed of acquisition by the technology in existence at that time, it is now possible to create files larger than 100, ,000 pixels. In the future, emerging acquisition technology that uses a multisensor design and miniaturized microscope arrays, may markedly decrease acquisition time, to less than 1 minute. 6 Conversion The single uncompressed image file was cropped and enhanced for brightness, contrast, and sharpness in Adobe Photoshop (Adobe Systems, San Jose, CA). This large file was then converted into a single multiresolution pyramidal file called flashpix. This.jpeg compression file contains multiple levels of resolution equivalent to 40, 20, 10, 5, 2.5, 1.25, and magnification. Roxio/MGI software (Roxio Inc, Santa Clara, CA) was used for the conversion to flashpix. Serving and Viewing The compressed flashpix virtual slide files of approximately 100 mb each were loaded onto a Dell PowerEdge 4400 computer (Dell, Round Rock, TX) running on a Linux operating system and Apache web server (Apache Digital, Durango, CO). Identifying information for each slide (slide number, diagnosis, and laboratory unit) was added to a MySQL database (MySQL AB, Uppsala, Sweden). Customized Perl CGI scripts were used to generate a table of contents for the virtual slides in each unit from the MySQL database and to deliver the images to the computer screen in html pages. Virtual slides were served with the MicroBrightField Virtual Slice system, which has as components MGI Zoom server software (Roxio) and a Java applet viewer imbedded in an web browser window (Fig 2). A zoom function in the viewer changes magnification by 2-fold with each click of the mouse by jumping from 1 resolution level to the next in the flashpix file. A click-and-drag function permits scanning of each magnification in an x y axis by streaming from the server to the viewer only the portion of the slide being viewed. Customized Perl CGI scripting is also used to direct a MicroBrightField java applet annotator to add arrow coordinates and text to the MySQL database, and subsequently to the slide when an html call is made by clicking on an annotation button (Fig 3A). A Perl CGI scripted side-by-side viewing function (Fig 3B) will be added to the Workshop laboratory site in This will allow comparison of cancer slides with normal, hyperplastic, and other cancer types. Once the slides and database information for the Workshop laboratory were prepared and added to the server, a Workshop laboratory website was generated from the database and made available to faculty before the course started. Because the Workshop site in Keystone does not have web access in each laboratory, Dell Inspiron 8500 laptop computers, running Microsoft XP (instead of Linux), were loaded with the flashpix virtual slide files and software (Apache server, the Perl scripted MySQL database, and MicroBright- Field Virtual Slice server and viewer system), allowing them to run as local-host web servers in the Workshop laboratories. In July 2001, one Workshop laboratory (male genitourinary tract) was equipped with a laptop computer containing virtual slides and a computer projector. The 5 other laboratories continued using traditional photomicrographs projected with a 2 2 carousel or Microsoft PowerPoint with a laptop and computer projector. Faculty in the male genitourinary tract laboratory used the virtual slides to demonstrate structures and lesions that students were expected to observe under a microscope. At the end of the laboratory sessions, the students were asked to complete a formative evaluation form that asked them to compare the teaching value of virtual slides in this laboratory versus the projected photomicrographs in the other laboratories. In July 2002, all of the 6 laboratories were equipped with virtual slides. RESULTS The evaluation data from July 2001, shown in Table 1, indicate that, compared to photomicrographs, virtual slides (1) were of higher image quality and resolution, (2) allowed the faculty to better point out cells and lesions that students were expected to see on the glass slide, and (3) enhanced the students ability to learn from glass slides at the microscope. Student comments in support of virtual slides included It really helped to have instructors find structures, I was better able to identify the specific features on my slide, This definitely helped with identifying lesions better than normal projector slides, and they helped (me) to know that I was looking at the correct structures. Some criticisms of the virtual slides are that they can make you lazy and that some students didn t look in detail at their own slides. In July 2002, based on the positive formative evaluations obtained in July 2001, all 6 Workshop laboratories were equipped for virtual slide projection. Although it was not possible to make comparisons again, positive comments about the use of virtual slides mirrored those of Similar to previous years, the Workshop laboratories were evaluated by the students as being very successful. For example, 91% of students strongly agreed that their knowledge of the histopathology of cancer increased. The mean ratings for both the lectures and the laboratories in 2002 are shown in Table 2. Note that the highest average student ratings in 2002 were for the laboratory subject matter (4.85) and quality of the slides (4.64). Additionally, the entire faculty agreed that the virtual microscope was a useful and successful addition to the laboratories. DISCUSSION The longevity of the AACR s annual Pathobiology of Cancer Workshop is but one indicator of its success. Since its inception in 1978, the Workshop has been 4 FIGURE 2. a. (top) Low power (5x) view of a virtual slide. 2b. (bottom) High power (40x) view of a virtual slide. Magnifications of 0.625x, 1.25x, 2.5x, 10x, and 20x are also possible. Each magnification can be moved in the x-y axis by clicking and dragging the image. 433

5 HUMAN PATHOLOGY Volume 34, No. 5 (May 2003) 434

6 VIRTUAL MICROSCOPE SLIDES (Dee et al) TABLE 1. Student Comparison of Photomicrographs and Virtual Slides in 2001 (n 84) Compared to photomicrographs, the virtual slides... Strongly Agree Agree About the Same Disagree Strongly Disagree Were of high image quality and resolution Enhanced the instructor s ability to point out cells and lesions that you were expected to see on the glass slide Enhanced your overall ability to learn from the glass slides at the microscope based on teaching in small groups with a student to faculty ratio of about 8:1, which necessarily involves active participation by all. The advent of new technologies has brought real opportunities to enhance the students learning experiences. Even after only one year of implementation, the virtual slides appear to have had a significant impact on the educational environment over the previously used approach. The microscope teaching strategies used in the Workshop laboratories, now augmented with virtual slides, are similar to those used in many medical student pathology and histology course laboratories. Thus, we believe that virtual slide technology could be implemented in these traditional laboratory venues to replace both 2 2 photomicrographs and videomicroscopy, 2 modalities commonly used to augment student learning in microscope laboratories and small groups. The advantages of virtual slides are summarized in the following paragraphs. As opposed to the selective field of view of photomicrographs, an entire facsimile of a slide can be projected, magnified, and moved on the x y axis. This allows students to orient themselves on their slides with respect to the slide being projected. Note that in Figure 2, the screen retains a whole mount of the glass slide and a thumbnail, with a navigation box indicating the area of the slide being viewed. This greatly assists the faculty in navigating the slide, and indicates to students where they should be looking on their own slides. To paraphrase one Workshop student, virtual slides are the digital equivalent of a multiheaded microscope for the entire class. Moreover, each student simultaneously has his or her own microscope. Compared with the virtual slides system, in videomicroscopy the video camera has a very small low-power field of view (usually giving no lower than the equivalent of 5) and there is no thumbnail with a navigation box. An additional useful feature of virtual slides is the ability to preselect fields of view and add arrows. This feature (Fig 3A) allows the faculty to locate difficultto-find structures (eg, Reed-Sternberg cells, tripolar mitotic figures) before a laboratory session, and then easily retrieve them in the laboratory session by clicking on an annotation button. Unlike videomicroscopy, virtual slides can be viewed side-by-side (Fig 3B), allowing simultaneous comparison of different pathologic lesions and normal with abnormal tissue. Finally, the image quality of the digital virtual slides is far superior to that of videomicroscope images. Virtual slides have applications in settings other than the traditional microscope laboratory. At the University of Iowa, both histology laboratories and pathology case analysis exercises are now taught in combined computer and microscope laboratories and in small group rooms equipped with both virtual slides and videomicroscopy equipment. 3,7,8 In these settings, students can view the glass teaching slides with the traditional microscope, with videomicroscopy, or as virtual slides on the computer. From these options, students markedly prefer using virtual slides to prepare for small group sessions, and pathology faculty and students alike prefer using virtual slides for small group discussions. Major advantages of virtual slides from the student s perspective include efficiency and accessibility. In terms of efficiency, all of the slides are accessible, in focus, and with proper lighting and condenser adjustment, at the click of a mouse. The slides can also be integrated into html pages with patient information, gross images, and radiological images, and linked to other web-based resources. All of these features markedly facilitate student presentation of microscopic and other findings in small groups. Accessibility of web-based virtual slides allows students to prepare outside of scheduled laboratory time in computer laboratories, in student communities equipped with computers, and at home with a high-speed Internet connection (cable or DSL). Students can also study together at the computer screen in groups of their own choosing anywhere and anytime. Faculty also benefit from virtual slide technology in that they can prepare for teaching over the web; project virtual slides in auditoriums, laboratories, and small groups; and better interact with students as they evaluate virtual slides on the computer or projector screen. Faculty can also easily download screen shots of virtual slides or link to the source code of individual virtual slides for use in other computer-based educational programs. Outside of laboratory and small group teaching, emerging virtual slide technology has significant potential for innovation wherever traditional microscopy and photomicrographs are used. These venues include webbased publishing of atlases, textbooks, and articles; con- 4 FIGURE 3. a. (top) Annotated virtual slide at 40x. The arrows point at Reed Sternberg cells. 3b. (bottom) Side-by-side view of papillary carcinoma and normal thyroid. The bottom frame appears when the Normal histology Thyroid link is activated. 435

7 HUMAN PATHOLOGY Volume 34, No. 5 (May 2003) TABLE 2. Student Evaluation of Lectures and Laboratories in 2002 (n 98) Mean The lecture subject matter was interesting and valuable The lecture subject matter was largely new to me The lectures put a lot of emphasis on general principles that are broadly applicable to cancer The laboratory subject matter was interesting and valuable The laboratory subject matter was largely new to me The quality of the slides was excellent NOTE. Student responses rated on a scale of 1 (strongly disagree), 2 (somewhat disagree), 3 (neutral), 4 (somewhat agree), and 5 (strongly agree). tinuing medical education, computer-assisted case simulations and adaptive learning exercises; proficiency testing and certifying examinations; and telepathology. 6 Many of these venues for virtual slides were recently presented, displayed, and discussed at the Inaugural Symposium on Virtual Slides at the Medical University of South Carolina, Charleston, SC, November 2002 (more information available at musc.edu/vslide). A National Library of Medicine Information Systems Grant has allowed us to develop a public access database of histology and pathology virtual slides, the Virtual Slidebox of Histopathology (see path.uiowa.edu/virtualslidebox/ for more information). The Virtual Slidebox currently contains more than 600 virtual slides, including many from the AACR Workshop as well as many contributed by educators at other institutions. Initially, the content of the Virtual Slidebox relied on a core list of morphological entities of disease published in HUMAN PATHOLOGY in Public web access to slides in the database is available to course directors via an html link, allowing them to display virtual slides in their own html web pages. Also, course directors can request that virtual slides be mailed as tiff or flashpix files via removable media to serve from their own server. Because most of the originally acquired files are in a tiff file format, they can be edited and adapted by others for use in a wide variety of multiresolution pyramidal file server and viewer configurations. In summary, the evaluation of virtual slides in the Pathobiology of Cancer Workshop laboratories indicated that students were better able to grasp the morphological features of cancer in their glass slide sets when instructors used virtual slides rather than traditional photomicrographs to guide them. Although the initial cost of equipment and software for creating and delivering virtual slides is quite high, and industry standards have not yet crystallized, we believe that this new technology has the potential to revolutionize the way we teach and learn from microscopic images. Our hope SD is that virtual slide technology will be promoted and implemented in all of the various venues for microscopy, thus stimulating healthy competition among virtual slide developers as they continue to improve the quality of their virtual slide scanners, servers, and viewers. The end result should be that virtual slide technology rapidly becomes effective, efficient, and affordable for all educators involved in teaching microscopic anatomy and pathology. Acknowledgement. The authors gratefully acknowledge the support of the following Workshop faculty members who in 2002 helped create the AACR virtual slide laboratories and implemented them in their laboratory sessions: Jeffrey M. Arbeit, MD Stephen Baird, MD Michael B. Cohen, MD Carlyne Cool, MD Barry R. De Young, MD Fred R. Dee, MD Bette K. DeMasters, MD Roy A. Jensen, MD John M. Lehman, PhD Robert L. Low, MD, PhD Stewart Sell, MD David S. Strayer, PhD, MD Ann D. Thor, MD Frederic M. Waldman, PhD, MD Jianzhou Wang, MD, PhD University of California, San Francisco VA Medical Center, San Diego University of Iowa Health Care University of Colorado Health University of Iowa Health Care University of Iowa Health Care University of Colorado Health Vanderbilt University Medical Center Brody School of Medicine East Carolina University University of Colorado Health Albany Medical College Thomas Jefferson University Medical Center University of Oklahoma Health University of California, San Francisco Cancer Center University of Oklahoma Health The authors also thank Nadine Lomakin, Program Administrator, American Association for Cancer Research, for collecting evaluation data and coordinating the preparation of Workshop laboratories for virtual slide projection. REFERENCES 1. Silage DA, Gil J: Digital image tiles: A method for the processing of large sections. J Microsc 138: , Westerkamp D, Gahm T: Non-distorted assemblage of the digital images of adjacent fields in histological sections. Anal Cell Pathol 5: , Harris T, Leaven T, Heidger P: Comparison of a virtual microscope versus a regular microscope laboratory for teaching histology. Anat Rec 265:10-14, Steinberg DM, Ali SZ: Application of virtual microscopy in clinical cytopathology. Diagn Cytopathol 25:389-96, Leong FJ, McGee JO: Automated complete slide digitization: A medium for simultaneous viewing by multiple pathologists. J Pathol 195: , Weinstein RS, Descour MR, Liang C: Telepathology overview: From concept to implementation. HUM PATHOL 32: , Dick (Dee) FR: Web-based virtual microscope laboratories. Pathol Educ 25:58-62, Heidger P, Dee FR, Consoer D: An integrated approach to teaching and testing in histology with real and virtual imaging. Anat Rec 269: , Dillman D, Torner R, Finken L: Core morphologic concepts of disease for second-year medical students. HUM PATHOL 29: ,