O c t o b e r 2 0 1 2 Digital Pathology
TABLE OF CONTENTS Abstract... 3 Abbreviations... 4 Introduction... 5 Market Size... 6 State of the Art... 7 Commercial Digital Pathology... 9 Drawbacks of the Current State of the Art... 10 The Way Forward... 12 Conclusion... 14 Reference... 15 Author Info... 16
Abstract Pathology quite literally means the the study of disease. Modern pathologists have a number of important functions, one of which is to analyze sliced sections of abnormal tissues. These tissues are stained, mounted on a glass slide and viewed under a microscope for diagnosing the disease. Digital pathology deals with aspects of acquisition, management and interpretation of pathology information generated from a digitized glass slide. Digital pathology is a rapidly growing field which has generated a lot of interest among the stakeholders, including hospitals, laboratories, pathologists, surgeons, research institutes, residents, medical device manufacturers, healthcare companies, and pharmaceutical companies, among others. Although it is a new field, technological breakthroughs are occurring at a rapid pace. Digital pathology attacks the central challenge facing pathology, in that the physical coupling of pathologists and the slides, which has stalled the increase in pathologist productivity, would no longer be a factor. Today, there are high-end slide scanners which can scan a standard slide at 200x magnification in less than two minutes, generating whole slide image data in the multiples of gigabytes. However, a great deal of technical bottlenecks remain, and companies involved in creating digital pathology solutions are challenged by the lack of adoption of digital pathology systems. This whitepaper tries to provide a comprehensive review of the digital pathology landscape, both from a clinical and a technological perspective. 3
Abbreviations Sl. No. 1 Acronyms (Page No.) In the Order of Appearance DP (5, 6, 7, 8, 9,10,12,14) Digital Pathology Full Form 2 FDA (6, 10, 14) Food and Drug Administration 3 ROI (7) Region of Interest 4 WSI (7,12) Whole Slide Images 5 LIS (8, 14) Laboratory Information Systems 6 OEM (9, 12) Original Equipment Manufacturer 7 IHC (10) Immunohistochemistry 8 ER (10) Estrogen Receptor 9 PR (10) Progesterone Receptor 10 IVD (14) In Vitro Diagnostics 4
Introduction DP has the potential to dramatically increase throughput, insure against damage to glass slides, reduce manual error, and industrialize the service. According to the Digital Pathology Association, digital pathology (DP) is defined as a dynamic, image-based environment that enables the acquisition, management and interpretation of pathology information generated from a digitized glass slide [1]. It can be further subdivided into virtual microscopy and telepathology. Virtual microscopy deals with the scanning, storing, transmitting and posting pathological slides on one or more computer systems. Telepathology deals with the analysis of the slides by one or more pathologists who are not physically present at the site of glass slide. DP attacks the central challenge facing pathology departments over the years the physical coupling of pathologists and slides, which has stalled the increase in pathologist productivity by retarding scalability. Digital pathology creates an environment where a pathologist can access whole slide images whenever and wherever he/she chooses, either through a computer system or even on a smartphone. It allows them to consult multiple senior pathologists at the same time and get their opinions without physically transporting the slide. It creates a vista of opportunities for educational and research institutes. It obviates the need for the physical availability of slides and a microscope for residents when a case is being taught. Images can be stored for decades without loss of quality. DP opens up possibility of creating pathology centres in remote locations where pathologists cannot be stationed. It also opens up avenues of consistency in slide scoring by removing ambiguity associated with human eye. DP has the potential to dramatically increase throughput, insure against damage to glass slides, improve productivity and efficiency, reduce manual error, intelligently assist pathologists, surgeons and hospital administrators to identify and rectify bottlenecks in laboratory workflow, improve timely treatment decisions and ensure better patient care. 5
The current DP market size is estimated to be around $150 million. It is expected to grow to $2 billion by 2020. About 1.5 billion histopathological slides are investigated manually each year in the US alone. Market Size According to GE Healthcare, the current DP market size is estimated to be around $150 million. It is expected to grow to $2 billion by 2020 [2]. About 1.5 billion histopathological slides are investigated manually each year in the US. Currently, digital pathology is used predominantly by pharmaceutical companies which do toxicological studies, veterinary pathologists, and educational institutes. Academic medical centers, commercial labs and large independent pathology labs are initial adopters of digital pathology systems. However, it will play a big role in future healthcare, particularly once the FDA approves digital pathology systems for primary diagnosis. Some of the earliest DP adopters have been veterinary pathologists. This is because many of them work for pharmaceutical companies engaged in basic research and clinical trials which involve drug toxicology and oncological studies. Pathology is essential to successful drug development for these companies. Hence, there exists a convincing argument for these companies to adopt digital pathology systems sooner than classical pathology labs catering to the front end of the healthcare sector. These large pharmaceutical companies have established multisite global image networks for storing, retrieving and sharing animal and human pathology whole slide images. This point was emphasized by Dr. Steve Potts, CEO of Flagship Biosciences, in one of his blogs. He states, nearly all of the top 15 pharmaceutical companies have completed multisite-integration with digital pathology. The result of such conversions is that a pathologist working in one location can access slides at any other one or participate in an informal peer review or pathology working group. [3]. 6
Z-stacking or multi-plane scanning of a specimen is the technique whereby the scanner captures the complete slide at various focal planes which will enable the user to focus the slide to the plane of choice. State of the Art There is a rapid growth in the development of DP technology. A few years back, only static image transfer was available, where the person transmitting the image decided the region of interest of his/her choice. The operator at the transmitting end captures an image which he perceives to be relevant and transfers it to a consultant for review. To operate, this requires a digital camera, the necessary software to enable transmission over a network, and a network connection. Image size is small and shows only a limited portion of the entire specimen. Therefore, remote observers will only see what the originator has imaged. The consultant at a distant location cannot change the magnification and choose an ROI of his/her choice. While this was cheap and quite straightforward, it created a built-in bias by not showing the whole slide to the telepathologist. This was a decisive disadvantage which stymied the growth of digital pathology. This problem was addressed by the advent of slide scanners which can scan the entire slide and generate large whole slide images (WSI). Whole-slide images are very large static images. Even as it maintains the simplicity of static images, it eliminates the bias because the entire specimen is now available to the telepathologists for review and comments. But, just like static images, WSI lacks a multi-planar focusing capability. This can be a barrier when cytological or hematopathological slides are analyzed. This is because the cells usually reside in multiple focal planes, bringing the third dimension (z-axis) into play. In conventional practice, the pathologist updates his focus continuously as he moves from one field of vision to another. This isn t possible in simple whole slide imaging which captures images in just one focal plane. However, techniques to overcome this shortcoming are now available. Z-stacking, or multi-plane scanning of a specimen is the technique whereby the scanner captures the complete slide at various focal planes. This will enable the user to focus the slide to the plane of choice, i.e. the plane where the cells are in focus at that particular field of view. Even the image transfers through networks have been considerably improved. Bandwidth requirements are no longer considered a bottleneck because modern transfer protocols don t require the transfer of the entire image file. Only the ROIs being queried at any point are transferred. Self learning algorithms can also increase the speed of viewing by transferring the surrounding areas of an ROI being currently viewed anticipating the user s next move. Although whole slide imaging (WSI) became possible, the quality of the digital whole slide images was not comparable to analog 7
Current state of the art digital pathology systems seamlessly integrate scanners to the consultant s workstation and also to hospital servers and laboratory information systems, which in turn can be connected to the internet. images. Even if the images were good, the size of the images (up to 15 Gb) made it impractical to be adopted on a large scale. The process of scanning one whole slide was also painstakingly slow. These pain points are being addressed by image compression algorithms and continuous incremental engineering breakthroughs. The DP workflow can be automated to a high degree. The automation process starts with tissue processing and continues with slide preparation, staining, ranking according to priority, and scanning the whole slide. A scanner can capture the whole slide and convert it into an image which then can be stored, retrieved and viewed independently by any number of pathologists at whatever resolution they choose. This brings unprecedented autonomy and freedom to the pathologists who are no longer tied by the unyielding laboratory workflow process of the 20th century. With improving camera hardware, computer processing power, storage capability, data transfer speed, image compression techniques and new software development, not only whole slide imaging is a possibility nowadays, but the use of digital images in pathology is being realized for a large spectrum of functions. This functionality can be at multiple levels, ranging from simple tasks such as storing and retrieving old case files, maintaining a database of interesting cases to the mid-level tasks of education, telepathology, multiple collaborative consultation, proficiency testing, securing data, and finally higher-level tasks of image processing, quantification, data-mining and the development of intuitive, self-learning artificial intelligence tools. Current state of the art digital pathology systems seamlessly integrate scanners to the consultant s workstation and also to hospital servers and laboratory information systems. These can then be connected to the internet and the details accessed from anywhere around the world on a variety of platforms. Figure 1: Typical DP Workflow 8
GE has collaborated with the University of Pittsburgh Medical Center and formed a company, Omnyx, which focuses exclusively on digital pathology. This collaboration is GE's first such ownership deal with an academic research institution. Commercial Digital Pathology Jondavid Klipp, the founder and editor of Laboratory Economics, states that roughly 500 commercial labs, pathology groups and academic medical centres in the USA have a digital pathology system/scanner in place [4]. The USA, in turn, accounts for approximately 60% of the digital pathology market. The market share of different OEMs is given below. Bioimagene has since been acquired by Ventana, which is fully owned by Roche. Figure 2: Market share of commercial Digital Pathology vendors (Source: Laboratory Economics Digital Pathology Trends Survey July 2012 n=174) [5] Philips and GE have entered the DP space, which many in the industry see as a validation of concept. The industry believes this will push the market toward higher adoption of DP systems. While Philips is working with NEC Japan to develop end-to-end DP solutions, GE has collaborated in a 50-50 partnership with the University of Pittsburgh Medical Center and formed a company, Omnyx, which focuses exclusively on digital pathology. This collaboration is GE's first such ownership deal with an academic research institution [6]. Omnyx has since designed an integrated digital pathology solution for research use. The services offered by the commercial products include rapid, high-resolution sample scanners, a hosting service to store the digital slides, and software to help read the slides on a computer monitor. Although there are minor differences, the throughput of currently available slide scanners is around 40 slides per hour (for a standard slide at 20x magnification). The cost of DP systems can vary from around $30,000 for desktop sized slide scanners to upward of $1 million for a comprehensive digital pathology system complete with a few state of the art scanners, processors, servers, networking infrastructure, computers and at least two highresolution, 20-inch monitors (similar to monitors used in radiology). 9
Drawbacks of the Current State of the Art Cost, inertia toward adopting a new technology and image data size are the top three reasons to delay the decision to plunge into digital pathology. Although the usage of digital pathology systems is high among veterinary pathologists, big pharmaceutical companies, academic and research institutions, it has not cracked the clinical diagnostics market. Digital pathology utilization for primary clinical diagnostics in United States is extremely limited, currently less than 1% of all slides [4]. The barriers to adoption of digital pathology systems is described in Figure 3. Cost, inertia toward adopting a new technology and image data size are the top three reasons to delay the decision to plunge into digital pathology. The biggest barrier to more clinical use is the cost of scanning digital slides, which doesn t eliminate the need to prepare glass slides first. Even those departments and labs which are not exactly short of resources choose to invest in workflow mission-critical equipment like automated tissue processors, slide stainers, glass coverslippers and immunostainers (for IHC), and postpone making decisions on DP systems installations. Figure 3: Barriers to adoption of Digital Pathology Systems (Source: Laboratory Economics July 2012) [5] The FDA has classified digital pathology systems for primary diagnosis as a class 3 medical device, and no company has yet obtained clearance for their system for diagnostic use of digital pathology. Aperio and Ventana are the only companies to have FDA approval for Her2 immuno histochemistry scoring. Ventana additionally has FDA approval for ER, PR, p53 and Ki 67 scoring algorithms for breast cancer. While this gives the vendors access to the periphery of the digital pathology market, it is the FDA approval for use of DP systems for primary diagnosis that the solution providers are eagerly looking forward to. Amanda Lowe, principal of Digital Pathology Consultants, believes the lack of it is a bigger issue than what the survey suggests. There s no doubt that if the FDA would gold stamp digital pathology on the clinical diagnostic end, it would help significantly, she says [4]. 10
Coming to the speed of the systems, a single plane scan can take up to five minutes, depending on the area of the specimen. With the higher-resolution scanning that will optimize the high magnification (40x objective scanning) image for cytology specimens, which means multi-planar focusing (z-stacking), scanning can take 30 40 minutes Z-stacking requires the storage of file sizes that are multiples of the already large single- plane whole-slide images (several hundred Mb each). Systems currently in use to transfer z-stacks over networks can be very slow, and can therefore lead to user fatigue and frustration. 11
Alternative techniques to z-stacking have emerged and they involve using scanners that allow for multi-plane scanning with stitching of the best focused image at each tile into the final singleplane file. The Way Forward The problems associated with z-stacking will be solved with solutions like faster scanners, faster computers, faster networks and increased storage capacities. Alternative techniques to z-stacking have emerged, and they involve using scanners that allow for multiplane scanning with stitching of the best focused image at each tile into the final single-plane file. In this method, the scanning time may not improve over z-stacking. This is because each plane still needs to be scanned individually. However, the final stitched file uses the memory and transfer requirements of a single-plane scan. A number of digital pathology solution providers are working on this technique, as the focus of the whole-slide image has been shown to be significantly improved. A study aimed at defining differences in performance of observers using 2-dimensional vs. 3-dimensional cytology images failed to show any difference in the diagnostic accuracy, but it indeed highlighted uncertainty and frustration on the part of observers at not being able to resolve out-of-focus cells on the 2-dimensional WSIs. For increasing the adoption of digital pathology systems, a number of pain points need to be addressed. A look at the results of the survey conducted by Laboratory Economics [5] makes it clear that expense is the most important barrier for widespread adoption of digital pathology. Two fifths of the respondents felt that DP solutions are too costly. While the costs will come down when economies of scale kick in, the OEMs should proactively look at cost reduction as a fundamental strategy to improve their business. One strategy would be to use cloud-based technologies. This will reduce the cost of storage and alleviate the need for standalone dedicated server and network systems. Houston-based software company Smart Imaging Technologies aspires to bring cloud computing and web applications for whole slide imaging to digital pathology. Their Simagis Live solution lets the user upload, store, share and retrieve whole slide images on the cloud. Another reason for the high cost of DP systems is the cost of the scanner. The digital pathology package usually includes a slide scanner, an image server and software for handling and viewing digital slides, and will cost upwards of $150K for a typical installation. This price tag may be attainable for a large hospital chain with sizeable budgets and a dedicated IT staff. However, a large number of pathology labs are small setups with minimal staffing. They have no capital budgets or IT support. It is impossible for them to splurge on large IT investments or technical support staff. These labs typically expect technology that requires minimal maintenance or support needs. To overcome this difficulty, many firms, including the market leaders, have started offering slide scanning services whereby the company will scan and digitize glass slides for a fee. Some companies are also specializing in cloud 12
hosting and web applications. A desirable cloud hosting and web application solution should have seamless uploading capability, be vendor-neutral, operate on open standards, be secure, and be affordable to small labs. Serving the small business and resident pathologists market economically is the logical progression for digital pathology system manufacturers. A number of innovative start-ups have managed to bring down the cost of scanning equipment, thereby bringing it within the affordability range of a small-scale, small-volume lab. They manufacture slide scanners which can be installed on the pathologist s desktop, with which s/he can scan interesting cases for discussion or educational purposes. Residents and students can use such systems to capture important and unique cases. By using modern manufacturing process and technologies, they have managed to bring the hardware price under $30K. 13
Conclusion The interest shown by industry leaders like Roche, Siemens, Philips and GE points to their confidence in the future and the potential of this field Digital pathology is a promising field within In-Vitro Diagnostics (IVD). Its case use extends to pathologists, labs, hospitals, teaching and research institutes, veterinarians, pharmaceutical companies and toxicologists. However, digital pathology has not found favour among clinical pathologists, and there are various reasons for that. Until now, early adopters of digital pathology systems have been the pharmaceutical companies who need to do toxicological studies, and veterinary pathologists. The biggest roadblock seems to be that digital pathology systems as a primary diagnostic tool have not been approved by the FDA. Further, the cost of the systems and the image sizes generated by DP systems are contributing to the slow uptake by clinical labs and hospitals. There is a general fear that they will clog hospital networks, and compatibility issues with the existing HIS and LIS are also contributing to the low adoption of DP systems. However, the technical challenges have been addressed head-on by the system vendors, and the interest shown by industry leaders like Roche, Siemens, Philips and GE points to their confidence that these issues will be sorted out in due time. In the meantime, companies should focus on the unmet needs for digital pathology by small labs, researchers and students who can t splurge on capital intensive systems. 14
Reference [1] Digital Pathology Association, Glossary of terms. Retrieved online on 27 th July 2012 from https://digitalpathologyassociation.org/glossary-of-terms_1 [2] Bio Spectrum Asia Interview with Mr Rajiv Enand, senior vice president, business development, Omnyx, GE Healthcare. Retrieved online on 27 th July 2012 from http://archive.biospectrumasia.com/content/040112oth1783 7.asp [4]Bruce Friedman Putting Some Numbers to Digital Pathology Adoption Trends by Pathologists Retrieved online on 27 th July 2012 from http://labsoftnews.typepad.com/lab_soft_news/2010/06/adopti on-trends-in-digital-pathology.html [5] Laboratory Economics; July 2012. [3] Steve Potts. Vets versus MDs in adoption rates Retrieved online on 27 th July 2012 from http://www.flagshipbio.com/regulatory/glp-whole-slideimaging/vets-versus-mds-in-adoption-rates/ [6] Novel Collaborative Relationships Emerge on the Medtech Landscape, Retrieved online on 27 th July 2012 from http://www.mddionline.com/article/novel-collaborativerelationships-emerge-medtech-landscape 15
Author Info Dr. Shyam Thangaraju is a medical doctor who has completed his Masters in Medical Science and Technology from the Indian Institute of Technology, Kharagpur. He works for the HCL practice team specializing in the medical device vertical. 16
Hello, I m from HCL s Engineering and R&D Services. We enable technology led organizations to go to market with innovative products and solutions. We partner with our customers in building world class products and creating associated solution delivery ecosystems to help bring market leadership. We develop engineering products, solutions and platforms across Aerospace and Defense, Automotive, Consumer Electronics, Software, Online, Industrial Manufacturing, Medical Devices, Networking & Telecom, Office Automation, Semiconductor and Servers & Storage for our customers. For more details contact eootb@hcl.com Follow us on twitter: http://twitter.com/hclers Visit our blog: http://ers.hclblogs.com/ Visit our website: http://www.hcltech.com/engineering-services/ About HCL About HCL Technologies HCL Technologies is a leading global IT services company, working with clients in the areas that impact and redefine the core of their businesses. Since its inception into the global landscape after its IPO in 1999, HCL focuses on transformational outsourcing, underlined by innovation and value creation, and offers integrated portfolio of services including software-led IT solutions, remote infrastructure management, engineering and R&D services and BPO. HCL leverages its extensive global offshore infrastructure and network of offices in 26 countries to provide holistic, multi-service delivery in key industry verticals including Financial Services, Manufacturing, Consumer Services, Public Services and Healthcare. HCL takes pride in its philosophy of 'Employees First, Customers Second' which empowers our 85,335 transformers to create a real value for the customers. HCL Technologies, along with its subsidiaries, has reported consolidated revenues of US$ 4.3 billion (Rs. 22417 crores), as on TTM ended Sep 30 '12. For more information, please visit www.hcltech.com About HCL Enterprise HCL is a $6.2 billion leading global technology and IT enterprise comprising two companies listed in India - HCL Technologies and HCL Infosystems. Founded in 1976, HCL is one of India's original IT garage start-ups. A pioneer of modern computing, HCL is a global transformational enterprise today. Its range of offerings includes product engineering, custom & package applications, BPO, IT infrastructure services, IT hardware, systems integration, and distribution of information and communications technology (ICT) products across a wide range of focused industry verticals. The HCL team consists of over 90,000 professionals of diverse nationalities, who operate from 31 countries including over 500 points of presence in India. HCL has partnerships with several leading global 1000 firms, including leading IT and technology firms. For more information, please visit www.hcl.com