1. 1st Congress of the International
Academy of Digital Pathology
August 3-5, 2011
Loews Hôtel Le Concorde
Québec City, Canada
Program and
Abstract book
http://iadphomepage.org
ORGANIZERS:
MASSACHUSETTS
GENERAL HOSPITAL

3. TABLE OF CONTENTS
Welcome address 2
Congress organization 3
Sponsors and Exhibitors 4
General information 5-6
Schedule
Wednesday, August 3 9-12
Thursday, August 4 13-15
Friday, August 5 16-17
Abstracts
Wednesday, August 3 20-32
Thursday, August 4 35-43
Friday, August 5 43-47
1|Page

4. Welcome to the 1st Congress of the International Academy of
Digital Pathology
Dear colleagues,
We are delighted to have you here to participate and share in the 1st Congress of the International Academy of
Digital Pathology organized by the International academy of Digital Pathology (IADP) in Quebec, Canada and
partners.
We are living in the era of great scientific achievements. Like other scientific areas, pathology is transforming fast in
this age of innovation. Digital pathology is not only here to stay, it has already gone way beyond the scanners and
external hard drive. Nevertheless, we still have a lot of challenges. One challenge is to standardize the systems after
taking various aspects into consideration, be it common image format, data privacy issues, business process re-
engineering, regulations, or medico- legal issues.
The other big challenge is to spread the benefits of digital pathology to a larger part of the humanity by including
the developing world into its fold.
With warm regards,
Dr.Yukako Yagi, USA Dr. Marcial Rojo, Spain
President Vice-President
International Academy of Digital International Academy of Digital
Pathology (IADP) Pathology (IADP)
2|Page

5. International Academy of Digital Pathology
Board Members
Yukako Yagi, Boston, USA (President)
Marcial Garcia Rojo, Ciudad Real, Spain (Vice president)
Gian Kayser, Freiburg, Germany
Klaus Kayser, Berlin, Germany
Jacques Klossa, Paris, France
Arvydas Laurinavicius, Vilnius, Lithuania
Vincenzo Della Mea, Udine, Italy
Janusz Szymas, Poznan, Poland
Yasunari Tsuchihashi, Kyoto, Japan
Guillermo Tearney, Boston, USA
Bernard Têtu, Quebec City, Canada
Treasurers
John Gilbertson,USA
Andrew Bajko,USA
Secretaries
Maristela Onozato, USA
Pinky Bautista, USA
International Advisor
Ronald Weinstein,USA
3|Page

6. SPONSORS and EXHIBITORS
DIAMOND SPONSOR
PLATINUM SPONSOR
GOLD SPONSORS
SILVER SPONSORS
EXHIBITORS
4|Page

7. GENERAL INFORMATION
uu
Registration
The congress registration is located at Cullen at the same floor where the lectures and exhibits
are held. Registration desk will be opened every day.
August 3 – 07:30 – 05:00
August 4 – 07:30 – 12:00
August 5 – 08:00 – 12:00
Badges
Registered participants will receive their badges. The badge serves as a ticket to conference
sessions, refreshment breaks and lunches. Participants are kindly requested to wear their
badges when attending the congress.
Presentation
Please submit your PowerPoint presentation file at least 2- 3 hours before your scheduled
presentation in the speaker’s ready room in Pilot.
Poster Session
The poster session will take place at Boduas-krieghoff1 on August 3, 17:30 -18:30. Authors
should put up their posters on August 3, between 10- 12 Noon. The posters should stay in the
same location until lunch break on Friday August 5, 2011 for viewing during breaks.
Wireless Internet Connection
Wireless internet connection is provided in the congress venue. Use the following code to
access the wireless internet connection: QUELO01 COJADU.
5|Page

8. GENERAL INFORMATION
Refreshment breaks and Lunches
Refreshments will be served at the Foyer and lunches will be served at Place Montcalm.
Welcome reception
Hors d'oeuvres reception during the poster session on August 3, 17:30 -18:30
Exhibition
Technical exhibits will be held at the Borduas-krieghoff1 and at the Foyer of the congress venue.
Exhibition hours:
August 3 9:20 – 18:30
August 4 8:00 – 18:30
August 5 8:00 – 12:00
6|Page

9. 7|Page

10. Attend the company’s presentation on August 4, 17:10 -17:20
Shun Doi
e-Pathologist
8|Page

11. WEDNESDAY, August 3 Room
08:00 – 17:00 REGISTRATION Cullen
07:00-08:00 Foyer
BREAKFAST
8:00 - 08:30 Suzor-kote,Krieghoff2
OPENING PLENARY
08:00 - 08:15
Dr. Yukako Yagi & Dr. Bernard Têtu
Opening talks
08:15 - 08:30
THE QUEBEC MINISTER OF HEALTH
Short talk
PLENARY SESSION Suzor-kote,Krieghoff2
Chair: Dr. Yukako Yagi, USA, & Dr. Marcial Rojo,Spain
08:30 - 09:00
Bernard Têtu, MD
Laval University, Canada
The challenges of implementing a “patient-oriented” telepathology network; the Eastern Quebec telepathology project
experience
09:00 -09:20
Michael Riben, MD
MD Anderson, USA
Non-Destructive Digital Pathology using Light-CT ™ on Fresh and Fixed Tissue: Initial Experience for Clinical and Research
Applications
09:20-09:30 Foyer, Borduas
SHORT BREAK
PANEL DISCUSSION: STANDARDIZATION IN DIGITAL PATHOLOGY Suzor-kote,Krieghoff2
Chair: Dr. Klaus Kayser & Dr. Bernard Têtu
09:30 -10:00
Marcial Rojo, MD
Hospital de Ciudad Real, Spain
Standardization Efforts of Digital Pathology in Europe
10:00 - 10:30
Yukako Yagi,PhD
Massachusetts General Hospital, USA
Toward Standardization: Color and Image Quality Validation in Whole Slide Imaging (WSI)
10:30 - 11:00
Michael Meissner,PhD
Co-Chair of DICOM WG26,USA
DICOM Working Group Activity
11:00 - 11:30
Andrew Evans ,MD,PhD
College of American Pathologists, CAP
Validating Whole Slide Imaging for Diagnostic Purposes in Pathology: Recommendations of the College of American
Pathologists (CAP) Pathology and Laboratory Quality
11:30-12:00 DISCUSSION/ Q&A
9|Page

12. 12:00 – 13:00 Place Montcalm
LUNCH BREAK
13:00-14:10 Suzor-kote,Krieghoff
PLENARY SESSION
Chairs: Dr. Yukako Yagi, USA, & Dr. Marcial Rojo,Spain
13:00 - 13:40
Ronald Weinstein, MD Keynote speaker
The University of Arizona,USA
Z-Axis Challenges in Whole Slide Imaging (WSI) Telepathology
13:40 - 14:10
Guillermo Tearney, MD,PhD
Massachusetts General Hospital,USA
Endoscopic microscopy: bridging the radiology-pathology divide
14:10-15:10 Suzor-kote,Krieghoff
SCIENTIFIC SESSION
Chairs: Dr. Gian Kayser,Germany & Dr. Arvydas Laurinavicius,Lithuania
14:10 -14:30
Judit Zubovits, MD
Sunnybrook Health Science Center, Canada
Proof of principle: Colocalization of pan cytokeratins (AE1/AE3), pan cytokeratin (PCK26), and cytokeratin 8/18 using an
Integrated Sequential Staining and Imaging Device
14:30 - 14:50
Catherine Bor,MD
Francois Baclesse Center, France
Benefits of a two resolution strategy for analyzing virtual slides of immunostained tumors
14:50 - 15:10
Christopher Malon,PhD
NEC Laboratories America,USA
Mitotic Figure Recognition: Agreement Among Pathologists and Computerized Detector
15:10 – 15:30 Foyer, Borduas
BREAK
15:30 – 17:10
SCIENTIFIC SESSION Suzor-kote,Krieghoff
Chairs: Dr. Janus Szymas, Poland, & Dr. Robert Osamura,Japan
15:30 - 15:50
Junya Fukuoka, MD
Toyama University, Japan
Application of digital pathology to create pathology assisting software to standardize the diagnosis of interstitial pneumonia
15:50 - 16:10
Yasunari Tsuchihashi, MD
Louis Pasteur Centre for Medical Research, Japan
Possibilities of Digital Navigation Systems for Pathology Diagnosis
16:10 -16:30
Yukako Yagi, PhD
Massachusetts General Hospital, USA
Evaluation of Ultra High WSI Viewing System
10 | P a g e

13. 16:30 - 16:50
Shigeatsu Yoshioka,PhD
Japan
Ultra High Speed WSI Viewing System
16:50 - 17:10
Yasuhiro Fukunaga,MS
Olympus, Inc., Japan
Spectral sensing method for practical use
17:10 - 17:30
Shinsuke Tani,MS
Olympus, Inc., Japan
Color Standardization System Implementing Estimation Method for Absorption Spectra of Dye
17:30 – 18:30 Borduas-Krieghoff1
POSTER SESSION
Chair: Dr. Bruce Levy
Jason Hipp, MD
University of Michigan, Department of Pathology, USA
Optimization of detection of complex cancer morphology using the SIVQ pattern recognition algorithm
Rajyasree Emmadi , MD
University of Illinois at Chicago, USA
Tissue refractive index as an objective and quantitative measure of pathologic processes
Rajyasree Emmadi , MD
University of Illinois at Chicago, USA
Mid-Infrared Spectroscopic Imaging for Breast Tissue Histopathology: Towards ‘Stainless Staining
Rajyasree Emmadi , MD
University of Illinois at Chicago, USA
Medical School Pathology education supplemented with web-based virtual microscopy
Sushmita Mukherjee, MD
Cornell University, USA
Toward an annotated digital Multiphoton Microscopy (MPM) histology atlas of fresh human bladder biopsies for intra-
cystoscopy guidance in bladder cancer diagnosis
Yasuhiro Fukunaga
Olympus, Inc,Japan
Spectral sensing method for practical use
Shinsuke Tani
Olympus, Inc.,Japan
Color Standardization System Implementing Estimation Method for Absorption Spectra of Dye
Pinky A. Bautista, PhD
Massachusetts General Hospital, USA
Multispectral enhancement towards digital staining
Peter Kragel, MD
Brody School of Medicine, East Carolina University, USA
A Pathology Imagery Interpretability Rating Scale for Virtual Microscopy
Anurag Sharma, MS
NEC Laboratories, USA
Balancing image quality and compression factor of special stains whole slide images
11 | P a g e

14. Anurag Sharma, MS
NEC Laboratories, USA
Digital pathology hidden challenges
Maristela L. Onozato,MD,PhD
Massachusetts General Hospital,USA
Automated 3D-reconstruction of histological sections
Maristela L. Onozato, MD,PhD
Massachusetts General Hospital, USA
Automated sectioning machine for paraffin blocks
12 | P a g e

15. THURSDAY, August 4 Room
08:00 – 17:00 REGISTRATION Cullen
07:00 – 08:00 Foyer
BREAKFAST
08:00 – 09:00 Suzor-kote,Krieghoff2
PLENARY SESSION
Chairs: Dr. Yasunari Tsuchihashi & Dr.Andrew Evans
08:00 – 08:30
Philippe Camparo, MD
Hôpital Foch, Suresnes, France
Whole slide imaging and expert analysis in adenocarcinoma of the prostate
08:30 – 09:00
Robert Osamura, MD
International University of Health and Welfare (IUHW) Mita Hospital, Japan
Digital Pathology in Asia
09:00 -10:00 Suzor-kote,Krieghoff2
SCIENTIFIC SESSION
chairs: Dr. Yasunari Tsuchihashi & Dr. Andrew Evans
09:00 - 09:20
Arvydas Laurinavicius,MD
National Centre of Pathology and Vilnius, Lithuania
Digital Image Analysis in Pathology: Benefits and Obligations
09:20 – 09:40
Janusz Szymas,MD
University of Medical Sciences, Poland
Web Microscope as a computer - based system for practical examination of dental students in oral pathology
09:40 – 10:00
Gian Kayser,MD
University of Freiburg, Germany
Virtual microscopy - a new tool in quality control and assurance of automated laboratory processes
10:00 – 10:20 Foyer
BREAK
10:20 – 12:00 Suzor-kote,Krieghoff2
SCIENTIFIC SESSION
Chairs: Dr. Philippe Camparo , France & Dr. Guillermo Tearney, USA
10:20 – 10:40
David McClintock,MD
Massachusetts General Hospital, USA
Using Computerized Workflow Simulations to Assess the Feasibility of Whole Slide Imaging Full Adoption in High Volume
Histology Laboratory
10:40 – 11:00
Manabu Fukumoto,MD,PhD
Tohoku University, Japan
The e-Pathologists Cancer Diagnosis Assistance System for Gastric Biopsy Tissues
13 | P a g e

16. 11:00 -11:20
Mari Mino-Kenudson, MD
Massachusetts General Hospital, USA
Role of 3D reconstruction in the classification of lung adenocarcinoma
11:20 - 11:40
Peter Gould, MD,FRCPC
Laval University, Canada
A comparison of digitized frozen section and smear preparations for intraoperative neurotelepathology
11:40 - 12:00
Slawomir Walkowski, MD
University of Medical Sciences, Poland
Histopathologic patterns of nervous system tumors based on computer vision methods and whole slide imaging (WSI)
12:00 – 13:00 Place Montcalm
LUNCH BREAK
13:00 – 14:10 Suzor-kote,Krieghoff2
PLENARY SESSION
Chairs: Dr. Yukako Yagi, USA & Dr. Ronald Weinstein,USA
13:00 - 13:40
Klaus Kayser,MD keynote speaker
UICC-TPCC, Institute of Pathology, Germany
How to Introduce virtual microscopy (VM) in routine diagnostic pathology: constraints, ideas and solutions
13:40 - 14:10
Katia Manova,PhD
Memorial Sloan-Kettering Cancer Center,USA
Molecular Cytology Core Facility- Efforts to Accelerate the Research Process
14:10 – 15:10 Suzor-kote,Krieghoff2
SCIENTIFIC SESSION
Chairs: Dr. Gian Kayser, Germany & Dr. Janus Szymas, Poland
14:10 -14:30
Nicolas Ellie, MD
CLCC Francois, CAEN,France
Automatic analysis of virtual slides to help in the determination of well established prognostic parameters in breast
carcinomas
14:30 - 14:50
Roy Lee, MD
Massachusetts General Hospital, USA
Automated quantification of liver steatosis by WSI based image analysis
14:50 - 15:10
Hatice Cinar-Akakin,PhD
The Ohio University, USA
Application of Multi-scale Filtering to Cell Nucleus Detection with Automated Scale Selection
15:10 -15:30 Foyer, Borduas
BREAK
15:30 – 16:50 Suzor-kote,Krieghoff2
SCIENTIFIC SESSION
Chairs: Dr. Katia Manova,USA & Dr. Jun Fukuoka
14 | P a g e

17. 15:30 - 15:50
Tokiya Abe, Phd
Keio University,Japan
Quantification of Liver Fibrosis by Whole Slide Image Analysis
15:50 - 16:10
Chamidu Atupelage, PhD
Tokyo Institute of Technology,Japan
Multifractal Feature Descriptor for Histopathology
16:10 - 16:30
Peter Kragel, MD
Brody School of Medicine at ECU,USA
A Pathology Imagery Interpretability Rating Scale for Virtual Microscopy
16:30 -16:50
Noriaki Hashimoto, MS
Tokyo Institute of Technology,Japan
Development of Image Quality Evaluation Method for Whole Slide Imaging
16:50 – 17:00 Foyer, Borduas
SHORT BREAK
17:00 – 18:30 Suzor-kote,Krieghoff2
COMPANY DISCUSSION FORUM
Chairs: Dr. Klaus Kayser, Germany, & Dr. Bernard Têtu, Canada
17:00 – 17:10
Pierre F. Le Fèvre
Aurora Interactive Ltd.
The difference between the "professional medical" part of digital pathology versus the "Infrastructure part"
17:10 – 17:20
Shun Doi
NEC Corporation,Japan
e- Pathologist
17:20 -17:30
Name:
Olympus, Inc.
Title:
17:30 -17:40
John Wellbank & Jose Castanon
Philips
Next Generation Digital Pathology
17:40 – 17:50
Philippe Nore
Roche Diagnostics
Roche Tissue Diagnostics’ Digital Pathology Strategy
17:50 -18:30
DISCUSSIONS/Q&A
15 | P a g e

18. FRIDAY, August 5 Room
08:00 - 12:00 REGISTRATION Cullen
07:00-08:00 Foyer, Borduas
BREAKFAST
08:00 – 09:00 Suzor-kote,Krieghoff2
PLENARY SESSION
Chairs: Dr. Paul Fontelo,USA & Dr. Essam Ayad,Egypt
08:00 - 08:30
Andrew Evans, MD, PhD
University of Health Networks (UHN), Toronto, Canada
Telepathology Based on Whole Slide Imaging for Primary Frozen Section Diagnosis: The University Health Network
Experience and Learned Along The Way
08:30 - 09:00
Jean –Paul Fortin, MD
Laval University, Canada
Promoting knowledge translation through telehealth evaluation: A strategic choice for telepathology projects
09:00 – 10:00 Suzor-kote,Krieghoff2
APPLICATION & IMPLEMENTATION OF DIGITAL PATHOLOGY
Chairs: Dr. Andrew Evans,Canada & Dr.Jean-Paul Fortin,Canada
09:00 - 09:20
Martin Yaffe, PhD
Sunnybrook Research Institute & University of Toronto, Canada
Applications of Digitized Whole-mount Histopathology
09:20 - 09:40
Kar-Ming Fung, MD, PhD
University of Oklahoma Health Sciences Center, USA
Whole Slide Images and Digital Media in Pathology Education, Testing and Practice: The Oklahoma Experience
09:40 - 10:00
Bernard Têtu,MD
Laval University, Canada
Laval University Telepathology System (demonstration at exhibit booth)
10:00 – 10:10 Foyer, Borduas
SHORT BREAK
10:10 -12:15 Suzor-kote,Krieghoff2
DIGITAL PATHOLOGY IN THE DEVELOPING COUNTRIES AND BEYOND
Chairs: Dr. Yukako Yagi,USA & Dr. Marcial Rojo, Spain
10:10 - 10:40
Paul Fontelo, MD
National Library of Medicine,USA
Digital Pathology – Implementation Challenges in Developing Countries Context
10:40 -11:10
Aliyah Sohani, MD
Massachusetts General Hospital,USA
Static Digital Pathology: A Model for a Regional Diagnostic and Educational Support Network for Pathologists in the
developing World
16 | P a g e

19. 11:10 -11:40
Essam Ayad, MD
Cairo University, Egypt
Virtual Microscopy Beyond the Pyramids, Application of WSI in Cairo University for E-education & Telepathology
11:40 - 12:10
Gian Kayser, MD
University of Freiburg, Germany
History, experiences, and perspectives of expert consultation in diagnostic surgical pathology
12:10 - 12:30
Marcial Rojo, MD
Hospital de Ciudad Real, Spain
Awards and Closing remarks
12:30 – 13:30 Foyer, Borduas
LUNCH/CONGRESS ADJOURNMENT
17 | P a g e

20. 18 | P a g e

21. Attend the company’s presentation on August 4, 17:00 – 17:10
Pierre F. Le Fèvre
The difference between the "professional medical" part of digital pathology versus the
"Infrastructure part"
19 | P a g e

22. The challenges of implementing a services that were unavailable for so many years ; 2) The
network must adapt to a changing situation occurring in
"patient-oriented" telepathology the course of the deployment, such as the moving and
network; the Eastern Quebec retirement of pathologists; 3) This unique network
telepathology project experience combining centers with or without laboratory requires the
development of innovative solutions (new report format,
Bernard Têtu gross station for large specimens and discussions with the
surgeon); 4) Pathologists used to send consultations by
RUIS-Laval University telepathology project mail must get used to send and read images; 5) The
network must be developed within each region because
Québec, Canada
University hospital are unable to absorb an increased
August 3, 8:30 – 9:00 workload; 6) Surgeons and pathology personnel must
Background: The Eastern Québec telepathology project develop new working methods; 7) Technicians in hospitals
was initiated by the Laval University Integrated University without pathology laboratory have limited experience with
Health Network (in french: RUIS: Réseau Universitaire histology techniques. Conclusion: The Eastern
Intégré en Santé). The RUIS was created by the Québec Québec telepathology network has been designed to
ministry of health as an authority of coordination and improve medical care to patients in this territory.
consensus which mandate is to encourage the integration However, the project team is confronted with major
of health cares, teaching and research activities in the challenges mostly related to change management and
institutions under the umbrella of the faculty of medicine. requiring innovative solutions.
This project is aimed at providing uniform diagnostic
telepathology services in a territory of 408,760 km2 with a
Non-Destructive Digital Pathology using
population of 1.7 millions inhabitants and which density, in
certain areas, is as low as 0.4 inhabitant/km2. Clinical Light-CT ™ on Fresh and Fixed Tissue:
context: Many surgeons in smaller community hospitals Initial Experience for Clinical and
must postpone surgeries requiring a frozen-section or to
Research Applications
transfer certain patients to regional hospitals because of
the lack of stable pathology coverage. Younger
Michael Riben
pathologists in early practice feel insecure and are often
reluctant to work alone because of the impossibility to MD Anderson, USA
rapidly obtain a second opinion from a colleague and can
August 3, 9:00 –9:20
hardly be absent without disturbing the surgical unit
organization. In this context, the ministry of heath and
Context: It is increasingly important to maximize the
Canada Health Infoway decided to financially support this
volume and quality of tissue obtained for both clinical and
innovative initiative with the objective of providing the
research use, particularly in the current push for delivering
population with uniform pathology services on the territory.
personalized cancer therapy. Given the expanding
The project is particularly timely because of the reports of
potential uses for collected tissue, minimizing consumption
two recent commissions of inquiry in Canada
while confirming the histologic diagnosis and assessing
recommending the deployment of telepathology to help
the quality of tissue obtained, both of which plague current
providing professional support and improve the quality of
methods such as frozen section or cytologic assessment,
cares in Canada. Scope of the project : this is a
is critical. To this end, we have begun testing a novel
“patient-oriented” project aimed at providing surgeons and
commercial device for full-field optical coherence
pathologists with frozen sections and second opinions
tomography called Light-CT™ which facilitates rapid non-
anywhere and anytime on the whole territory of the RUIS-
destructive histologic digital imaging of both fresh and
Laval University. Each hospital is equipped with a whole
fixed tissue. The technique is based on the principle of
slide scanner, a gross station, a videoconferencing device
white light interferometry and generates quick image slices
and with a viewer and an image sharing solution. The
in three-dimensions of the specimen. Our initial
project has been implemented in 21 sites, of which 6 are
experience includes evaluation of the ability to use the
devoid of pathology laboratory. Of the remaining 15 sites,
images for quality control and the potential for diagnostic
6 have one pathologist, 7 have 2 or more pathologists and
assessment. Methods: We obtained residual tissue from
two have no pathologist. Change management is the
gastrointestinal tract, breast, and skin for imaging. Full
major challenge encountered in the implementation of this
field and virtual sectioned slides of tissues at a variety of
network. Basically, we found that 1) Although the
depths were acquired using a 10x/0.3 numerical aperture
surgeons of the 6 hospitals devoid of pathology laboratory
immersion objective. Images were processed and
and the 2 devoid of pathologists are motivated by the
visualized in real-time. The imaged tissue was submitted
potential of the technology, they must get used to request
20 | P a g e

23. for traditional histology, after which comparison of the processes according to feedbacks we have gotten from
Light-CT™ image and the glass slide were undertaken. the pathologists. These results can be beneficial to
Results: Initial experience show architectural and improve the organization of digitizing centers, since they
morphologic similarity between the Light-CT image and can used to create models and protocols enabling
histologic preparations. Distinct morphologic criteria telepathology international network websites for
includes pixel intensity, patterns which correlate with teleconsultation services. In the next future, this work can
histologic features, and architectural variability. Together, also be applied in the documentation of pathology
these provide adequate information for a quality processes described in many standardization documents
determination. Image correlation with the stained H&E (DICOM, HL7, IHE).
section highlights image features with associated
histologic features including architecture, tumor invasion, Working Group 2 has been dedicated to Informatics
tissue reaction, and normal structures. Conclusions: Standards in Pathology, and promoting their application in
Light-CT™ represents a paradigm shift for histologic Pathology Departments in European institutions. The main
assessment of both fresh and fixed tissue. We have research is focused in: A) Advances in Integrating
shown that the images produced show great potential for Healthcare Enterprise (IHE)- Anatomic Pathology:
use-cases in which pattern recognition and architectural Anatomic Pathology Workflow (APW) within the hospital,
assessment are used to evaluate both for quality of tissue Anatomic Pathology Reporting to Public Health (ARPH)
obtained and diagnostic criteria. Future work includes and Anatomic Pathology Structured Report (APSR), based
assessing molecular markers after imaging. in HL7 Clinical Document Architecture (CDA). B) The
practical application of DICOM standard (medical image)
in Pathology. C) Image compression. D) Semantic
Standardization Efforts of Digital interoperability: PathLex and SNOMED CT in Pathology.
Pathology in Europe
Image compression research discussion has been
performed in collaboration with JPEG committee. As a
Marcial García Rojo
conclusion of this discussion, it seems that JPEG 2000
Hospital General Unviersitario de Ciudad Real. Spain image compression has proved to be a most efficient
August 3, 09 :30 – 10 :00 image compression standard, but is has not become
widely implemented. JPEG commitee is working in the
EURO-TELEPATH stands for “Telepathology Network in development of a new compression method, JPEG XR,
Europe”, a European COST Action mainly aimed at those digital photography applications
(http://www.conganat.org/eurotelepath/) that started in where JPEG 2000 penetration has been limited due to
2007 and will end in November 2011. This project is aimed complexity reasons, and they are also interested in
to foster collaboration between European research groups evaluating Advanced Image Coding (AIC) in medical
working in an adequate technological framework for the images.
management of multimedia electronic healthcare records,
in informatics applied to Anatomic Pathology, and, most
importantly in IT standards applied to digital medical
Toward Standardization: Color and
images, collaborate with international standardization Image Quality Validation in Whole Slide
bodies. Imaging (WSI)
Working Group 1 “Business modeling in Pathology”
Yukako Yagi, Noriaki Hashimoto, and Pinky Bautista
performed a comparison study between different notations
in business modeling (Business Process Modeling Department of Pathology, Harvard Medical School,
Notation – BPMN, Event Process Chain – EPC, and UML Massachusetts General Hospital, Boston, MA
Activity Diagram). They have concluded that BPMN is the
modeling notation which is clearer and more August 3, 10 :00 – 10 :30
understandable to pathologists. EPC is harder for
Context: Standardization of the image quality and the
pathologists to read and understand. It is probably
color displayed by digital slides are important aspects of
because EPC is not as popular as BPMN and UML AD, so
digital pathology implementation. While the most common
pathologists are not so familiar with the concept of event-
reason for the variations of color and image quality is the
driven process.
variance in the protocols and practices in the histology lab,
the image displayed can also be affected by variation in
Main pathology processes – Frozen Study, Formalin Fixed
capture parameters (for example, focus, illumination and
Specimen Study, Telepathology, Cytology, Autopsy – are
filters), image processing and display factors in the digital
now complete. Further work will be needed to modify the
systems themselves. It is difficult to identify which exactly
21 | P a g e

24. cause the problem. We have developed the methodology development will be slow and availability of such
to standardize color and evaluate image quality of any algorithms will not be as widespread as it could potentially
WSI. Design: Two types of calibration slides were be. DICOM WG26 continues its efforts and activities to
developed in our laboratory: color and image quality establish and enhance interoperability, most recently
calibration slides. These slides were scanned using targeting the adoption of DICOM Worklists to define
different scanners. The color information from the scanned procedures and procedure steps that allow devices in the
image of the color calibration slide was used to pathology space to communicate respective status. In this
characterize the color settings of the scanners. The session, we will provide an update on the ongoing
effectiveness of the proposed color standardization and activities in DICOM WG26 as well as seek ideas for
image quality evaluation algorithms was demonstrated additional topics that WG26 could/should tackle in the
using the image quality evaluation slide, which is a ideally future. While Informatics is not something pathologists are
stained embryo tissue slide. Results: The scanned very familiar with / commonly exposed to, the future of
images showed varying color appearance on the same pathology lies heavily in Pathology Informatics and
monitor. However, after the implementation of the color DICOM WG26 would benefit from more active end user
standardization algorithm the color variance was (pathologists) participation which we would like to improve.
minimized. Moreover, there was a good correlation
between the results of the proposed image quality This talk will give an overview of the accomplishments of
evaluation method with the evaluation results by human DICOM WG26, current activities, and potential future
eyes. We have developed a tentative protocol for color efforts. We believe educating about such progress and
standardization and image quality validation. getting more active participation by various parties is a key
Conclusions: Initial results of the standardization ingredient to moving Pathology into the Digital Age.
experiments are promising although there are still several
factors that we need to consider. We have been
conducting a series of research towards Standardization.
Validating Whole Slide Imaging for
However, to promote the importance to have standardized Diagnostic Purposes in Pathology:
color of image and high quality of image and also Recommendations of the College of
standardize validation protocol, the collaboration between
Academic, Industry, Governmental organization may be
American Pathologists (CAP) Pathology
necessary. and Laboratory Quality Center
Andrew J. Evans1, John H. Sinard2, Lisa A. Fatheree3,
DICOM Working Group Activity Walter H. Henricks4, Alexis B. Carter5, Lydia Contis6,
Bruce A. Beckwith7, , Christopher N. Otis8, James
Michael Meissner MacDonald3, Anil V. Parwani6, Liron Pantanowitz6
1
Co-Chair of Dicom WG26,USA Laboratory Medicine Program, University Health Network,
2
Toronto, ON, Canada; Department of Pathology, Yale
August 3, 10:30 – 11:00 3
University School of Medicine, New Haven, CT; College of
4
American Pathologists, Northfield, IL; Pathology and Laboratory
DICOM WG26 has successfully completed two 5
Medicine Institute, Cleveland Clinic, Cleveland, OH; Department
supplements, Supplement 122 targeting primarily LIS of Pathology and Laboratory Medicine, Emory University, Atlanta,
6 7
vendors to exchange specimen information and GA; North Shore Medical Center Salem Hospital, Salem, MA;
supplement 145 primarily targeting imaging vendors to Department of Pathology, University of Pittsburgh Medical Center,
8
exchange WSI information. Without any doubt, those two Pittsburgh, PA; Department of Pathology, Baystate Medical
are major milestones in the process of establishing a Center, Tufts University School of Medicine, Sprinfield, MA
framework for standardized digital Pathology Informatics. August 3, 11:00 – 11:30
With standards being established, it is now up to the
Introduction: There is an increasing interest in using
vendors to implement those but even more so up to the
whole slide imaging (WSI) for diagnostic purposes
pathology institutions to demand vendors to support of
(primary and/or consultation). An important question is
these standards so that the adoption can occur more
whether WSI can replace conventional light microscopy as
quickly. Having those standards is important but even if
the method by which pathologists review histologic
those were supported by most vendors todays,
sections and/or cytology slides to render diagnoses.
standardized Pathology Informatics needs more: i.e. most
Validation of WSI is crucial to ensure that diagnostic
devices today do not support device independent image
performance based on digitized slides is at least
information and support ICC profiles is critically important
equivalent to that of glass slides and light microscopy.
for the development of image algorithms that can be
Currently, there are no standardized guidelines regarding
applied across different vendors’ images. Without it,
validation of WSI for diagnostic use. Methods: The CAP
22 | P a g e

25. Pathology and Laboratory Quality Center convened a non- for cases in which specimens are imaged in a single focal
vendor panel from North America with expertise in digital plane. The participating pathologists had experience with
pathology. A total of 611 international publications that met both digital robotic telepathology and whole slide imaging
search term requirements were identified. Papers outside telepathology. The pathologists listed potential “problem”
the scope and related solely to technical components, surgical pathology cases where through-focal imaging
education and image analysis were subsequently (imaging of multiple planes in the Z-axis) could be
excluded. Publication year, author country, WSI clinical especially important.
application, number of observers and cases used,
validation method and outcome measurement were In this survey, the following types of cases were identified
recorded. Twenty two articles cleared the work group as potentially dependent on imaging multiple focal planes
requirements and were submitted for quality of evidence (Z-axis imaging). In surgical pathology, identifying and
grading by a contracted methodologist. Draft classifying microorganisms by conventional light
recommendations were developed from this microscopy can be challenging. This is especially so for
comprehensive literature review and panel consensus. certain bacteria such as Helicobacter pylori, because of
Final recommendations will be completed after a public their small size and shape, fungi, and parasites. The
comment period. Results and Conclusions : Validation of identification of viral inclusions in single focal plane digital
WSI is essential to ensure that pathologists using this images is challenging. Interpretation of renal biopsies for
technology for diagnostic purposes will consistently make transplantation rejection benefits from examining the
the same clinical interpretation as they would from glass distribution of lymphocytes in multiple planes. Also, in
slides and a light microscope. Validation should address surgical pathology, identifying crystals in single plane
both technical and interpretative components to digital images of biopsies can be difficult. In cancer cases,
demonstrate that the WSI system under review produces identification of mitotic figures in surgical pathology
acceptable digital slides for diagnostic interpretation. A sections, and differentiating normal from abnormal mitotic
pathologist must be involved in the validation process. In figures, can be problematic. There are many examples of
order to simulate actual clinical practice, a validation cytopathology specimens benefiting from multiple focal
sample set should be of a minimum size and include the plane imaging. Examples of cytology cases that might
spectrum of specimen types and diagnoses likely to be benefit from digital imaging in multiple planes include PAP
encountered with the intended clinical use. Measurable smears, and body fluids in which there might be a
outcomes should document diagnostic concordance tendency for cells to aggregate into balls. Cells in FNAs
(accuracy) between digital and glass slides for the same (fine needle aspirates) can also “ball up” into large
observer (intra-observer variability) with a reasonable aggregates greater in diameter than the thickness of a
washout period. Approved clinical use of WSI should be single Z-axis focal plane digital image. Identification of
limited to the conditions under which validation occurred. crystals and casts in urine specimens can be problematic
with single Z-axis plane viewing. Hematopathology
studies, using smears, can be challenging, because of the
Z-Axis Challenges in Whole Slide need for imaging in exquisite detail, often requiring oil-
Imaging (WSI) Telepathology immersion imaging with a through-focal series of images.
Interestingly, these “consensus” diagnostic accuracy
Ronald S. Weinstein, Anna R. Graham, Fangru Lian,
challenges often correspond to problem areas that are
and Achyut K. Bhattacharyya
underrepresented in the digital pathology literature. This
Department of Pathology, University of Arizona College of may reflect the well-known predilection of clinical
Medicine and Arizona Telemedicine Program, Tucson, Arizona investigators to study questions that are likely to yield a
August 3, 13 :00 – 13 :40 positive outcome. Before digital pathology is incorporated
into routine practice, guidelines for handling of potential
Surgical pathology diagnostic accuracy studies comparing “problem cases” should be incorporated into standard
telepathology (i.e., digital imaging, whole slide imaging, operating procedures.
virtual microscopy) and conventional light microscopy
often show slightly lower diagnostic accuracy for
telepathology. When the differences in diagnostic Proof of principle: Colocalization of pan
accuracy are not statistically significant, isolated cytokeratins (AE1/AE3), pan cytokeratin
diagnostic errors can still have serious implications for
(PCK26), and cytokeratin 8/18 using an
individual patients.
Integrated Sequential Staining and
We surveyed experienced telepathologists in order to Imaging Device
identify potential sources of diagnostic difficulty, especially
23 | P a g e

26. Judit Zubovits1, Kashan Shaikh3, Alex Corwin3, Dan
Wang4, Sean Dinn2, Gina Clarke4 Chris Peressotti4 ,
Zhengyu Pang2, Robert Filkins2, Martin J. Yaffe4
Benefits of a two resolution strategy for
1
analyzing virtual slides of
Department of Anatomic Pathology, Sunnybrook Health
2
Sciences Centre; Diagnostic and Biomedical Technologies, GE immunostained tumors
3
Global Research Center; Electrical Technologies & Systems, GE
4
Global Research Center; Ontario Institute for Cancer Research Catherine Bor1,2, Paulette Herlin1, Nicolas Elie1,
1
and Sunnybrook Research Institute Benoît Plancoulaine
August 3, 14 :10-14 :30 1
GRECAN, EA 1772, IFR146 ICORE, Caen University and
2
François Baclesse Cancer centre, Caen, France; Pathology
Background:Co-localization of biochemical and molecular
Department, François Baclesse Cancer centre, Caen, France.
markers at the cellular and subcellular levels remains an
active area of research and development within image August 3, 14 :30-14 :50
analysis. Cytokeratins (CK) are a family of about 20
different water-soluble proteins that form the cytoskeleton Background: The quantitative estimation of
of epithelial cells. All of them are used in the clinical lab to immunostained biomarkers on histological sections of
recognize epithelial cells. We demonstrate a unique tumors provides oncologists with prognostic and
optofluidic instrument and a multiplexing therapeutic indicators. To take into account the frequent
immunohistochemistry technique for sequentially labeling structural and functional heterogeneity of tumors, the
different proteins on the same section of breast cancer study of sampled representative whole sections of the
and illustrate its use in analyzing the co-localization tumors is mandatory. The fully automated analysis of high
between 3 fluorescent cytokeratin markers. Methods: resolution virtual slides offers a new opportunity in this
Antibodies: Pan cytokeratin (panCK) AE1/AE3 (Catalog# context. Nevertheless, due to their huge size, it is
M3515, DAKO, Carpinteria, CA); cytokeratin 8/18 necessary to define a swift and efficient strategy of
(CK8/18) (Clone K8.8 +DC 10), (Catalog# MS-1603-P0, analysis. Methods : The proposed approach includes the
Lab Vision Fremont, CA); panCK (clone PCK26) (Catalog# carefull subsampling of the high resolution image (0.5 or
C1801, Sigma, St. Louis, MS). A de-identified breast 0.25µm) up to the critical resolution of 4µm and the
cancer sample was collected from the archives of combined automatic analysis at both levels. The collection
Anatomic Pathology Department at Sunnybrook Health of data dealing with the whole tumor tissue is done at the
Sciences Centre. Biomarker multiplexing was performed low resolution level and on the whole image (mean and
using an automatic, sequential staining and imaging maximum density of the marker under study and
system. Images were registered, and the autofluorescence characterization of its distribution). The additional
signal removed. Co-localization was quantified on a pixel information concerning the cellular and subcellular
by pixel basis using Pearson’s correlation coefficient (r). A parameters is done at high resolution, taking advantage of
mean value of r was obtained over multiple regions of a cloning of the structures of interest or of a systematic
interest (ROIs) on the section. Results :Initially, to sampling of tissue details. The low resolution level allows
estimate operational stability, perfectly correlated samples to isolate the tissue compartments and immunostained
were obtained by imaging the same section but using markers of interest, while the high resolution level helps in
different exposure times. Due to noise in the image refining the segmentation done at low resolution. The
acquisition system, the r value was not 1 but 0.988 superimposition, on the full resolution image, of the
±0.001. A total of 14 ROIs each measuring 758 microns upscaled segmentation masks obtained at low resolution,
by 758 microns were acquired on a single section of allows pathologist’ visual control. The intersection of these
invasive ductal carcinoma of the breast. The correlations masks with the results of an interactive reference marking
between CK8/18 and panCK(PCK26), panCK(AE1/AE3) of a stereological grid, done at full resolution, leads to an
and panCK(PCK26), panCK(AE1/AE3) and CK8/18 were evaluation of the quality of the automatic analysis method
0.959 ± 0.016, 0.918 ± 0.03, 0.904 ± 0.029, respectively. which was implemented. Results : The two resolution
Since CK8/18 recognizes all simple epithelia including analysis which is illustrated through several examples
glandular epithelium, it is very likely that CK8/18 is the (quantification of nuclear markers and blood vessels) runs
primary CK in this breast cancer specimen, or all other on commercially available personal computers and allows
types of CKs are highly correlated to the expression of collection of data from the whole tumor sections in a very
CK8/18.Conclusion: Our work demonstrates a novel and short delay, compatible with the routine practice.
versatile system for co-localization studies that can be Conclusions : Thanks to this simple and easy to
easily adapted to work with existing commercially implement strategy, the swift analysis of various
available antibodies in a translational research setting. pathological virtual slides seems at hand.
24 | P a g e

27. Mitotic Figure Recognition: Agreement some strong biases exist (-.35, -.05, and .27). Thus, in the
weak agreement between A and B, pathologist C’s own
Among Pathologists and Computerized
vote often determined the majority result. Prediction would
Detector be much harder for an independent observer.
Conclusions : We have trained a computer to recognize
Christopher Malon2, Elena Brachtel5, Eric Cosatto2, Hans
mitotic figures, using the pathologists’ consensus. The
Peter Graf2, Atsushi Kurata3, Masahiko Kuroda3,
recall of this system for a high precision is within the
John S. Meyer4, Akira Saito1, Shulin Wu5,
bounds of the performance of the original pathologists, if
and Yukako Yagi5
they had to commit only to affirmative or negative
1 2
Innovative Service Solutions Division, NEC Corporation; Dept. decisions. Previous work has investigated agreement in
3
of Machine Learning, NEC Laboratories America; Dept. of mitotic grades, which are determined by the total number
4
Pathology, St. Luke's Hospital; Dept. of Molecular Pathology, of mitotic figures detected. We have evaluated pathologist
5
Tokyo Medical University; Dept. of Pathology, Massachusetts agreement on individual mitotic figure recognition, on an
General Hospital and Harvard Medical School unprecedented number of figure samples. In so doing, we
have exposed systematic differences in the kinds of
August 3, 14 :50-15 :10
patterns counted by different pathologists.
Background: We compare three pathologists’
References
assessments of mitotic figures and develop a system to
recognize mitosis using machine learning. Identification of 1. J. P. A. Baak, E. Gudlaugsson, I. Skaland, L. H. R. Guo, J. Klos, T.
H. Lende,H. Soiland, E. A. M. Janssen, and A. zur Hausen.
mitotic figures is a laborious task that is essential to Proliferation is the strongest prognosticator in node-negative breast
grading and prognosis for breast cancer. A mitotic count cancer: significance, error sources, alternatives, and comparison
with molecular prognostic markers. Breast Cancer Res. Treat.,
provides one of the three components of the Bloom– 115:241–254, 2009.
Richardson grade for invasive carcinoma [2]. Yet, 2. H. J. Bloom and W. W. Richardson. Histological grading and
according to several studies ([3], [4], and [1]), pathologist prognosis in breast cancer; a study of 1409 cases of which 359
have been followed for 15 years. Br. J. Cancer, 11:359–377, 1957.
agreement for mitotic grade is rather modest, with [3] 3. J. S. Meyer, C. Alvarez, C. Milikowski, N. Olson, I. Russo, J. Russo,
reporting agreement of κ =0.38. An automated mitosis A. Glass, B. A. Zehnbauer, K. Lister, and R. Parwaresch. Breast
carcinoma malignancy grading by Bloom-Richardson system vs
count could accelerate a pathologist’s work while proliferation index: reproducibility of grade and advantages of
improving reproducibility. Methods: We ask three proliferation index. Modern Pathology, 18:1067–1078, 2005.
pathologists to classify 4,204 sites as “mitosis,” “might or 4. P. Robbins, S. Pinder, N. de Klerk, H. Dawkins, J. Harvey, G.
Sterrett, I. Ellis, and C. Elston. Histological grading of breast
might not be mitosis,” or “not mitosis.” We compute carcinomas: a study of interobserver agreement. Hum. Pathol.,
statistics of agreement among these pathologists. Based 26(8):873–9, Aug 1995.
on majority–agreed sites, we train a computer to recognize
the mitotic figures. The computer uses image analysis Application of digital pathology to create
heuristics to extract candidates, then combines manually
designed features with features automatically learned by a pathology assisting software to
convolutional neural network, to make a final decision for standardize the diagnosis of interstitial
each candidate point, using a support vector machine. pneumonia.
Results: The machine’s performance falls within the
range of the humans’ performance if “maybe” labels must Junya Fukuoka, Manami Saito, Tomonori Tanaka,
be committed to be affirmative or negative: Kyoko Otani, Takashi Hori, Sayuri Nunomura
August 3, 15 :30-15 :50
Table 1. Prediction of test figures where two
pathologists agreed Background: Pathological diagnosis in the field of rare
disease usually requires special education and training.
Especially for the many non-neoplastic diseases, the
pathological diagnoses can only be judged by combination
of non-specific findings. Usually, the order of importance
and/or specificity for each finding is not well described in
the textbook, and experience of cases only can give
pathologists senses of judgment in each field. Considering
the amount of complexity and increase in number of
biopsy for upcoming personalized medicine and the
shortage of pathologists throughout the world, pathology
Among the three pathologists, we observe pairwise
assisting system to standardize pathological diagnosis is
agreements of κAB = .13, κAC = .44, and κBC = .39, but
highly needed. The area of interstitial lung disease (ILD) is
25 | P a g e

28. one of the most complex fields, and the accurate three major regional hospitals in Kyoto, Japan were
distinction of usual interstitial pneumonia (UIP), highly investigated in view of potential pitfalls for human errors
lethal disease, from other types of ILD is a critical but and efficiency of the diagnostic works. Possible digital
difficult role for the surgical pathologists. Material and pathology navigational systems for idealistic pathology
Methods: We have first examined the agreement of diagnosis were also extensively investigated. Results:
diagnosis for the 20 cases of interstitial lung disease Automatic review of such basic patient records as name,
among 29 pathologists including experts and non-experts age, sex, clinical records, etc, were possible in the routine
for pulmonary pathology. Second, to know the importance laboratory diagnosis of the regional hospitals but not in
of findings which leads to the diagnosis of UIP, a system their telepathology diagnostic systems. The lack of the
to record evaluating process of each finding before calling automatic review of the information constituted a potential
a specific pathological diagnosis was created using whole for human error. Review of digital radiology images of a
slide scanning images (WSI). In short, WSI of the same 20 patient was not available in two of the three regional
cases were analyzed by four pulmonary pathologists. The hospitals owing to limited function of their hospital
software was designated to evaluate 17 queries regarding information system (HIS). Comparison of digital radiology
histological findings using WSI of the 20 cases before images to digital macroscopic pathology images is
reaching to the final diagnosis. Region of interests (ROI) functionally possible but still not in common use.
images for positively scored findings were all captured for Comparison of digital macroscopic image at a specific
further discussion. Scoring data in each finding given by point of tissue sampling of an extirpated organ to its
four panelists were analyzed statistically using linear specific microscopic image is experimentally possible but
regression analysis, and the skewness coefficients (SC) currently not in practical use. Automatic detection of
for each finding toward the diagnosis of UIP were fragmented tissues on a slide and their realignment for a
calculated. Results and Discussions:The generalized certain diagnostic purpose as modified digital images is
kappa coefficient among 29 panelists was confirmed to be experimentally possible but still not in practical use.
low as 0.21. If it was limited to pulmonary experts, the Automatic detection of a certain tumour area in a given
kappa value was still not greatest (0.41). Among recorded digital microscopic image is now experimentally possible
17 findings by the software, 6 showed significant but still not in practical use.
positive/negative SC. The 6 findings include levels of
scarring and architectural destruction. Currently, we are Conclusions: It is clear that there are a lot of possibilities
creating the software that assists pathological diagnosis for digital navigational systems for pathology diagnosis to
for general surgical pathologists using the 6 findings. The enhance diagnostic efficiency or to ensure correct
utility of the software and digital pathology for the diagnosis. We need navigations for idealistic pathology
diagnostic processes and digital pathology will provide the
routine clinic will be discussed in the session as
solution.
well.
Evaluation of Ultra High Speed WSI
Possibilities of Digital Navigation
Viewing System
Systems for Pathology Diagnosis
Yukako Yagi1, Hiroshi Kyusojin2, Shigeatsu Yoshioka2,
Yasunari Tsuchihashi Maristela Onozato1, Eugene J Mark1, Matthew P Frosch1,
Department of Clinical Pathology Research, Louis Pasteur Centre David N Louis1
for Medical Research, Kyoto, Japan 1 2
Massachusetts General Hospital, Boston, MA, USA; Sony
August 3, 15 :50-16 :10 Corporation, Tokyo, Japan
Background: Pathology diagnosis is a series of August 3, 16 :10-16 :30
informational processes starting from a biopsy material or Context: The ability to digitize histopathology slides
from an extirpated surgical material to finally reach one automatically, rapidly and with high resolution has been
sentence of diagnosis. The diagnostic processes are pursued by numerous investigators around the world. One
known to harbor pitfalls for human errors and may include of the goals is to implement into the clinical practice in
monotonous laborious processes. In the age of digital Pathology. One of the pending issues is the “speed” of the
pathology we may be able to create various navigational entire process. The scanning time is faster and the
systems for ideal pathology diagnosis to ensure a correct performance of the scanner is more stable, however, the
diagnosis or enhance diagnostic efficiency. In the present viewing speed is still not satisfactory for most users. We
study possibilities of digital navigations in pathology have evaluated a new whole slide imaging viewing station
diagnosis are investigated. Methods: Pathology focusing on accessing speed, not only the speed to
diagnostic work flows of pathology laboratories of the manipulate the Whole Slide Images (WSI) but also the
26 | P a g e

29. speed to complete the tasks in many situations such as compression rate. Results: Due to the effect of high-
primary diagnosis, conference, teaching, etc. Design: speed processor and cache hit rate improvement using
PlayStation®3 and wireless controllers were adapted to movement prediction prefetch, most of operations are
our study because it can manipulate huge graphic data reflected by the display in 1/60s. Even in the case of
without any stress locally and also over the network. We cache miss, browsing latency is several times shorter than
have evaluated the system at three situations at the existing viewers. Moreover, applying game controller
Department of Pathology of MGH. 1. Simulation of brings several advantages including tireless panning and
signing-out cases, 2 Discussion tool at a consensus simultaneous multiuser operation. Conclusions: By
conference, 3. Usage at a teaching course. Prior to the using PS3®, our viewer eliminated the negative effects of
trials, we analyzed the required functions of the Graphic- pathology digitalization and derived the benefits. It can be
User Interface (GUI) and Human Interface for each the tool that improves the efficiency of daily operations.
purpose and slightly different GUI was made to fit each
purpose ideally. Results: The system had a remarkable Acknowledgment: The authors thank Department of
performance. The points of evaluation were 1. user- Pathology, Massachusetts General Hospital, Boston, MA, USA for
friendliness, 2. less frustration, 3 fit to the collaboration
purpose.Pathologists were able to use the system
comfortably after 0-15 min training. Pathologists who
played the game regularly at home required 0-3 minutes
Spectral sensing method for practical
training. Pathologists who never played the game required use
10-15 min training session. There were no complaints
regarding the speed. Most pathologists were satisfied with Yasuhiro Fukunaga, Saori Shimizu, Kensuke Ishii, and
the functionality, usability and speed of the system. The Kosei Tamiya
most difficult situation was to simulate signing-out of
Olympus, Inc.,Japan
cases.Conclusions: The preliminary results of our Ultra
High Speed WSI Viewing System were promising. The August 3, 16 50:10-17 :10
speed of viewing the images demonstrated the possibility Background: Digital color correction technology for image
to use WSI in the daily practice in a near future. We analysis and/or diagnosis is developed. Prior researchers
continue to evaluate the system for further improvements. have used monochrome camera with VariSpecTM tunable
imaging filter or field/frame sequential color camera.
Acknowledgment: The authors thank Department of However,it takes a long time to capture a number of
Pathology, Massachusetts General Hospital, Boston, MA, spectral images in series. Methods:The objective of this
USA for the collaboration work is to develop a spectral sensing method for practical
use. A light beam is divided into two by half mirror. One
light beam is for RGB camera and the other is for spectral
Ultra High Speed WSI Viewing System
sensing unit. The spectrum can be sensed at the same
time of capturing image by RGB camera. An area of
Shigeatsu Yoshioka1, Yukako Yagi2, Naoki Tagami1,
spectral sensing is smaller than image and larger than
Hiroshi Kyusojin1, Masashi Kimoto1, Yoichi Mizutani1,
1 1 picture element of RGB camera. Result: We applied the
Kiyoshi Osato , Hiroaki Yada
spectral sensing unit for color correction of stained tissue
1 2
Sony Corporation, Tokyo, Japan; Harvard Medical School and images. 20 spectrums of specimens were enough for color
Massachusetts General Hospital, Boston, MA, USA correction of whole slide image. Each spectrum was taken
within 235msec. We found it corrected the color variation
August 3, 16 :30-16 :50
of H&E images successfully. Conclusions: We proposed
a spectral sensing method suitable for digital color
Background: Over the years, we have gained real-time
correction for image analysis. We will evaluate
processing technology for minimum latency operation in
effectiveness of the proposed method for diagnosis with
game area. Among its many perceived applications to
Whole Slide Imaging system.
different areas, this technology will be especially useful for
digital pathology where it will allow the handling of huge
image data without delay. Methods: In this work, we Color Standardization System
applied PS3® to digital pathology viewer for whole-slide
images to evaluate the potential benefit. A wide variety of Implementing Estimation Method for
images were used for the evaluation of viewing speed and Absorption Spectra of Dye
usability – surgical biopsy, needle biopsy, mucosal biopsy,
polypectomy, cytology, H&E stain, IHC stain, etc. In terms Shinsuke Tani,Kensuke Ishi, Munenori Fukunishi
of data size, the largest one is about 2GB at 1/10
Olympus, Inc.,Japan
27 | P a g e

30. August 3, 17 :10-17 :30 Background: Image analysis algorithms, coupled with
maturing digital whole slide imaging (WSI) technology,
Content : Color standardization is indispensable for
holds promise to provide tools for morphometric
implementing “tele-pathology” and “computer assisted
quantitation in surgical pathology. However,
diagnosis” for clinical uses, because the color variations
implementation of such strategies will require development
affect the performance of digital image analysis as well as
and optimization of pattern recognition algorithms
confusion of pathologists. These color variations are
adaptable to diseases showing complex architectural
caused by not only device conditions but also staining
features and cytologic atypias. One such example is
methods, agent, environment, and so on. We developed
urothelial carcinoma (UC), of which the aggressive
color standardization based on multi-spectral techniques.
micropapillary variant (MPUC), an aggressive variant of
In our previous method, however, the workflow is not
UC which is frequently under-recognized causing
practical for pathologists, because it is necessary for them
diagnostic difficulties. Herein, we demonstrate the
to measure the absorption spectra using training samples
potential of a recently described pattern recognition
stained by pure dye. We propose the advanced method to
algorithm and its application to this challenging use case.
improve the accuracy and flexibility of color
Methods: We have recently reported SIVQ (Spatially
standardization without training samples. Method : We
Invariant Vector Quantization), an algorithm that uses ring
develop the method with a multi-spectral sensor module
vector predicates for pattern recognition (Hipp and Cheng,
and software installed into a digital microscope system.
2010). However, the relative contributions of key SIVQ
The algorithm is consists of the following procedure: 1) we
ring parameters have not been fully characterized.
acquire both the RGB image and the multi-spectral signals
Consequently, we systematically tested SIVQ ring
from multiple points on the specimens. 2) We estimate the
parameters for detection of micropapillary nests in fields of
absorption spectra of dye from the multi-spectral signals
a classic case of MPUC by comparing pattern match
based on statistical method. 3) We implement color
quality scores at pixels inside and outside of a pathologist-
standardization using the estimated results. Experimental
determined “ground truth,” using the receiver operating
results : Without any training sample, we successfully
characteristic (ROC) curve. Results: To standardize ring
estimate the absorption spectra of dye and standardize
parameter optimization, we tested various ring diameters,
colors of H&E images. We evaluate the effectiveness of
number of subrings, and inter-ring rotational “wobble”
our proposed method at the viewpoint of image feature
angles. First, we modulated the number of sub-rings
detection for digital diagnostic assistance. Conclusion :
(skipping every other ring) from 0 to max (diameter in
For color standardization, we proposed an advanced
pixels-1), finding incrementally increased AUC
method suitable for pathological workflow. We
successfully standardized the color variation of H&E performance (from 0.66 to 0.86). Secondly, increasing ring
images by applying the method to digital microscope vector diameter (3-25 pixels) demonstrated initial improvement
systems. through 11 pixels, then degradation in performance,
identifying an optimal ring size of 11 (max AUC 0.82). In
contrast, adjusting the angle of inter-ring “wobble” from
minimum to maximum (1-180 degrees) showed little effect
POSTER SESSION on the AUCs (<0.01 variation in AUC). Conclusions:
Optimization of SIVQ can yield impressive performance for
August 3, 17:30 18:30 detection of complex tumor architectural features. Using a
novel iterative discovery workflow applied to this use case
Optimization of detection of complex of MPUC tumor nest detection, we found that maximal
subrings showed better ROCs, identified an optimal ring
cancer morphology using the SIVQ
diameter, and identified minimal contribution of inter-ring
pattern recognition algorithm “wobble” to performance. This strategy constitutes the
first description of an algorithm capable of histologic
Jason Hipp1, Steven Christopher Smith1, Jerome Cheng1, identification of MPUC and provides a model workflow
1 2 2 broadly adaptable for future applications.
Scott Tomlins , James Monaco , Anant Madabhushi ,
1 1
Priya Kunju , Ulysses J. Balis
1
University of Michigan, Department of Pathology
M4233A Medical Science I 1301 Catherine, Ann Arbor, Michigan
2
48109-0602; Rutgers The State University of New Jersey
Department of Biomedical Engineering 599 Taylor Road
Piscataway, NJ
28 | P a g e

31. Tissue refractive index as an objective different biopsies. Conclusions: Our data demonstrate
that the refractive index distribution of tissue is a valuable
and quantitative measure of pathologic intrinsic marker of disease and can set the basis for a new
processes generation of computer-assisted, label-free
histopathology, to enable earlier disease detection, more
Zhou Wang1, Rajyasree Emmadi2, Krishnarao Tangella3, accurate diagnosis, and high-sensitivity screening.
Andre Balla1, Shamira Sridharan1 and Gabriel Popescu1
1
Department of Electrical and Computer Engineering, University Mid-Infrared Spectroscopic Imaging for
of Illinois Urbana- Champaign (UIUC) andThe Beckman
Institute for Advanced Science and Technology Breast Tissue Histopathology:
2
Department of Pathology, University of Illinois at Chicago (UIC) Towards ‘Stainless Staining’
3
Provena United Samaritans Medical Center
Walsh MJ1, Mayerich D1, Wiley EL, Emmadi R,2
2 1
Background: Traditional tissue examination and Kajdacsy-Balla A, Bhargava R .
pathology diagnosis comprises light microscopic 1
Department of Electrical and Computer Engineering, University
evaluation of formalin-fixed paraffin-embedded (FFPE) of Illinois Urbana- Champaign (UIUC) andThe Beckman
sections with various chemical stains. This is, however, a Institute for Advanced Science and Technology
fairly subjective process with inter and intra-observer
2
variability. Quantitative phase imaging (QPI) of unstained Department of Pathology, University of Illinois at Chicago (UIC)
FFPE tissue sections provides objective, label-free, highly
sensitive and quantitative data based on the intrinsic Background: Histopathology is the gold standard for
tissue refractive index. We developed Spatial Light disease diagnosis. Current histopathological techniques
Interference Microscopy (SLIM), a new white-light QPI use a panel of special stains and immunohistochemistry
method that combines Zernike’s phase contrast (IHC) to assess tissue architecture, determine cell types
microscopy and Gabor’s holography. Using SLIM we present and to classify cancers. Mid-Infrared (IR)
imaged entire unstained prostate tissues with a side-by- spectroscopic imaging is a novel approach to derive
side comparison with adjacent sections of H&E stained chemical images from tissues based on their inherent
slides. Methods: Eleven prostate biopsies from 9 patients biochemistry. Methods: Mid-IR images were obtained
were imaged with both SLIM and traditional light from over 200 individual patients using breast tissue
microscopy. We utilized SLIM to analyze 4 µm sections of microarrays. Serial sections were stained with a panel of
FFPE prostate tissues that had been de-paraffinized and 13 routinely used special stains and IHC stains. A
placed in xylene. Three successive slices were stained modified Bayesian classifier was built to assign image
with H&E and immunohistochemical stains using pixels to the correct cell types and Artificial Neural
antibodies against Cytokeratin 34 beta E12 (high Networks (ANN) to replicate staining. Using Mid-IR
molecular weight CK903) and Alpha methylacyl-CoA- imaging coupled with the modified Bayesian classifier it
racemase (AMACR, p504s) and imaged with the same was possible to segment breast tissue into the main 8-cell
microscope (10x objective) via the bright field channel types of breast tissue from a single unstained tissue
equipped with a color camera. For each biopsy, section. Results: The sensitivity and specificity as
pathologist identified regions of normal and cancer were measured by average Area Under the Curve (AUC) were
designated the gold standard and compared to the phase very high (AUC=0.9). Mid-IR imaging coupled with ANN
shift variance data obtained by SLIM. Results: The demonstrated that it was possible to accurately reproduce
spatially resolved scattering map obtained by SLIM the staining of the panel of stains, all in a single unstained
showed very good correlation with the designated benign slide. Conclusions: Mid-IR imaging coupled with
and malignant areas. Regions of high variance (short Bayesian classification and ANN could potentially be a
scattering mean free path) corresponded to the darker very valuable tool as an adjunct to current
staining associated with tumor in H&E sections. Our histopathological procedures, with the ability to take a
findings indicate that prostate cancer renders tissue more single unstained tissue section and give a decision on the
inhomogeneous, making it more strongly scattering than cell types present and also to replicate staining patterns.
adjacent benign tissue. These findings were further This approach could be particularly advantageous where
confirmed by anisotropy factor measurements wherein limited histological and cytological specimen is available
malignant tissues consistently exhibited higher g values. A for analysis. Moreover, it is amenable to quantitative
mode vs fluctuation contrast histogram constructed from analysis of each component. This novel approach
the SLIM data separates prostate cancer from normal with promises to revolutionize and expand the role of the
100% accuracy as tested on 100 tissue regions from 11 pathologists in both research and tissue diagnosis.
29 | P a g e