Impact case study database
Leeds Virtual Microscope: A new tool for pathologists diagnosing cancer
1. Summary of the impact
With digital pathology, doctors can diagnose cancer from high-resolution scans of biopsies. The Leeds Virtual Microscope (LVM), with its patented user interface, has enabled pathologists to diagnose patient cases more quickly than from glass slides and was the first digital pathology software capable of running on the ultra-high definition displays that pathologists need for such gigantic images (biopsy ‘slides’ are typically 10 billion pixels). The LVM research has changed pathology practice in one of the UK’s largest hospitals and has shaped several companies’ R&D programmes. The dissemination of LVM technology through Roche is now changing the way cancer is diagnosed worldwide.
2. Underpinning research
Prof Darren Treanor (University of Leeds, Faculty of Medicine and Health, 2006 to present) and Prof Roy Ruddle (University of Leeds, School of Computing, 2006 to present) have collaborated by combining their clinical and computing expertise respectively to improve virtual microscopy. Dr Rhys Thomas (University of Leeds, School of Computing, 2009-2016) and Prof Rebecca Randell (University of Leeds, School of Healthcare, 2009-2019) have supported the project with development of the LVM software, the design and conduct of the user evaluations.
A. Identifying the need for a better digital pathology viewer
Historically, pathologists viewed digital pathology slides on low-resolution monitors (e.g., 1024x768 pixels), and took 60% longer to diagnose cancer from a digital slide compared with viewing glass slides through a conventional microscope. There were two key reasons for this: (a) the monitors had too few pixels, so pathologists could see only 17% of the tissue area compared to the microscope, and (b) the user interface was very inefficient.
Our research to address those issues began by designing software to display digital slides on a 54 megapixel ‘Powerwall’ display. An evaluation demonstrated a major improvement over commercial digital slide systems, because pathologists made diagnoses as quickly with the LVM as with a conventional (glass slide) microscope [ 1].
B. Design of novel software to make clinical diagnoses from digital slides
Major funding from the National Institute for Health Research (NIHR; GBP690,000; 2009-2012) allowed us to design, implement and evaluate two further versions of the LVM providing a similar performance benefit running on a desktop PC with a multi-monitor ultra-high definition display. After some fundamental research [ 2], the LVM Version 2 [ 3] allowed the diagnosis of single-slide cases (gastro-intestinal biopsies and skin cancer), which together make up half of the slide workload in major cancer units.
LVM Version 3 was enhanced to diagnose complex cancer cases (with 12 – 25 slides, comprising 25% of total lab workload) [ 4], based on a patented “case world” layout [ 5], and won the 2014 Yorkshire & Humber NHS Innovation Award for Medical Devices and Diagnostics.
C. Evaluation of the LVM in a clinical setting
The LVM was evaluated in a series of controlled user experiments with consultant pathologists. While they made diagnoses more slowly with Version 2 of the LVM than with a conventional microscope, the difference was not statistically significant [ 3]. Version 3 allowed pathologists to make diagnoses as quickly as with a conventional microscope [ 4].
The LVM’s user interfaces [ 6] won the 2016 Best Paper Award in the ACM Transactions on Computer-Human Interaction, with the journal’s editor stating that this “ article stands out because it puts into practice—and challenges—accepted design principles for the navigation of such gigapixel images, against the backdrop of real work by medical experts”.
Follow-on funding
The evaluations led to commercial interest and the award of further grants from the EPSRC-funded Medical Technologies Innovation and Knowledge Centre (2012-2014; GBP130,000; EP/J017620/1), an EPSRC Impact Accelerator Award (2014-2015; GBP40,000; EP/K503836/1), and the Yorkshire & Humber NHS Deanery (2015-2016; GBP110,000). These grants allowed the development of essential extra functionality for the Leeds Teaching Hospitals NHS Trust (LTHT) adoption and Roche commercialisation.
3. References to the research
- Treanor, D., Jordan-Owers, N, Hodrien, J., Quirke, P., & Ruddle, R. A. (2009). Virtual reality Powerwall versus conventional microscope for viewing pathology slides: an experimental comparison. Histopathology, 5, 294-300. DOI: 10.1111/j.1365-2559.2009.03389.x
Evaluation with pathologists of LVM version 1, on a 54 megapixel Powerwall display. The results led to major (GBP690,000) research funding from NIHR.
- Ruddle, R. A., Thomas, R. G., Randell, R. S., Quirke, P., & Treanor, D. (2015). Performance and interaction behaviour during visual search on large, high-resolution displays. Information Visualization, 14, 137-147. DOI: 10.1177/1473871613500978
Research with ordinary participants (students, not doctors) which informed the LVM’s design.
- Randell, R., Ruddle, R. A., Mello-Thoms, C., Thomas, R., Quirke, P., & Treanor, D. (2013). Virtual reality microscope versus conventional microscope on time to diagnosis: An experimental study. Histopathology, 62, 351-358. DOI: 10.1111/j.1365-2559.2012.04323.x
Evaluation with pathologists of LVM version 2 with single-slide patient cases, on high-resolution (11 megapixel) desktop displays.
- Randell, R., Ruddle, R. A., Thomas, R. G., Mello-Thoms, C., & Treanor, D. (2014). Diagnosis of major cancer resection specimens with virtual slides: Impact of a novel digital pathology workstation. Human Pathology, 45, 2101-2106. DOI: 10.1016/j.humpath.2014.06.017
Evaluation with pathologists of LVM version 3 with large (12 – 18 slide) patient cases, on high-resolution medical-grade displays from Barco. The results led to three further grants (from EPSRC and NHS) and paved the way for commercialisation with Roche.
- Ruddle, R. A. & Treanor. D. (2015). Patent for virtual microscopy (USA US8970618).
Protects the LVM’s main intellectual property and, therefore, Roche’s uPath digital pathology software as well.
- Ruddle, R. A., Thomas, R. G., Randell, R., Quirke, P., & Treanor, D. (2016). The design and evaluation of interfaces for navigating gigapixel images in digital pathology. ACM Transactions on Computer-Human Interaction, 23(1), Article No. 5. DOI: 10.1145/2834117
This won the ACM ToCHI 2016 Best Paper Award, redefining design principles for overview-and-detail navigation interfaces and describing details of the design of LVM version 3.
4. Details of the impact
The LVM has produced three types of impact in three areas.
A. New digital pathology products
Software: The LVM research produced intellectual property spanning novel software [ 3,4,6], know-how and patents (US8970618; EP3489733). That intellectual property was assigned in 2016 to Roche-Ventana [ A] for a value that cannot be disclosed. Roche is a top 3 laboratory vendor (turnover GBP45 billion, 94,000 employees), with commercial distribution in over 100 countries and established relationships with anatomic pathology laboratories with over 50% market share worldwide [ B].
The LVM was the basis of a fundamental redesign of the entire Roche digital pathology system in 2016-2018 to provide users with “ an innovative and intuitive way of interacting with their digital pathology cases” [ B]. Treanor and Ruddle acted as consultants during the prototyping and detailed design phases for the new software (2016-2017) and were part of its beta-testing team (2018). Since the software’s commercial launch as the uPath digital pathology enterprise software in 2019 [ C], Roche have issued over 100 active uPath licences in Australia, Europe, the USA and other countries [ B].
Scanner hardware: FFEI (a British Small & Medium-sized Enterprise) are the Original Equipment Manufacturer (OEM) developer and manufacturer of the Roche DP 200 scanner, which is an integral part of the Roche’s uPath system and compatible with the LVM. In 2012, FFEI started discussions with Leeds about the LVM, under a non-disclosure agreement. The “ market established by LVM//uPath” helped FFEI secure internal and external investment to fund new imaging technology developments, upgrade R&D and production facilities, and retrain staff, driving their future business strategy [ D].
FFEI have supplied WSI devices to over 27 countries in North and South America, Europe, the Middle East and Africa, and the Asia-Pacific region. From 2016-2019 their revenues from medical imaging grew from 20% to 50% of total revenue (GBP12,000,000 per annum), and FFEI expect this trend to continue and include “ direct-to-market products as a diversification of FFEI's current OEM business model” [ D].
B. Influence pathology practice
The LVM underpinned pathology going fully digital in 2018 at the Leeds Teaching Hospitals NHS Trust (LTHT). LTHT is a major cancer centre and tertiary referral pathology laboratory serving a population of 3 million people, and England’s second largest acute hospital trust.
After pilot evaluations during the LVM development [ 3,4], LTHT chose the LVM as the digital pathology platform for a 15,000-slide validation study ( https://doi.org/10.1111/his.13403) that directly led to national Royal College Guidelines for the pathology profession (Jan 2018; https://www.rcpath.org/resourceLibrary/best-practice-recommendations-for-implementing-digital-pathology-pdf.html).
When LTHT pathology went digital, the LVM was rolled-out to the entire department. Benefits include efficiency (“ really speeded up interpretation and turnaround times of some of the most complex cases”), understanding of disease “ much clearer appreciation of low power distribution of the interstitial abnormalities”), reducing delays for patients (“ more timely delivery of second opinions between hospital sites”), and organisational change (“ allowed us to reframe our thinking around future Cellular Pathology strategy for the region”) [ E].
C. Shaping research & development programmes
As well as Roche and FFEI, the LVM also shaped the R&D programmes of two other companies under non-disclosure agreements.
Display hardware: Barco dominates the global medical display market ( https://www.marketsandmarkets.com/ResearchInsight/medical-display-market.asp). From 2012, the LVM team worked with Barco and gave privileged access to research results. Barco gained confidence that “ ultra-high definition medical displays made sense commercially” [ F], invested in developing a new product to double display resolution (to the 12 megapixel Coronis Unit) that thousands of doctors now benefit from, and in 2017 started a four-year project to develop a new collection of high-resolution pathology displays. Barco draws parallels between the MIT Media Lab and the way the LVM team “ had a visionary idea … quickly assembled a real working example, and sat people in front of it to see what it looked like” before “deploy[ing] it in a hospital to see how it should be improved” [ F].
Pathology software: The LVM influenced the Swedish multinational Sectra Ltd, whose products are used in 2,000 hospitals and clinics around the world. The LVM provided Sectra with “ two foundational insights”: (i) “ it was possible to develop a digital viewer with performance matching diagnosis with a conventional microscope, which was under much debate at the time”, and (ii) there was “ great potential” for digital diagnosis to go “beyond what was possible in the traditional microscope”. This “ had a great positive influence on the Sectra business decision to invest and start a branch in pathology” [ G].
The LVM’s success cemented Leeds as the go-to place for digital pathology R&D, evidenced by the Leeds-led National Pathology Imaging Collaborative (NPIC) [ H] (GBP30,000,000 Innovate UK/ GBP11,000,000 industry funding and Leeds Centre for Doctoral Training (GBP6,000,000 UKRI funding) in Artificial Intelligence for Medical Diagnosis and Care [ I]. They include Roche, FFEI, Sectra and NHS England as partners.
5. Sources to corroborate the impact
A. Press release about the sale of the of LVM technology to Roche ( https://www.leeds.ac.uk/news/article/3828/sale_of_virtual_microscope_technology; 23rd Feb 2016).
B. Letter about LVM acquisition, and design and licencing of the new Roche uPath about digital pathology enterprise software, from the Digital Pathology Life Cycle Leader, Ventana Medical Systems (23rd Oct 2020). Ventana is a member of the Roche Group of companies, and part of Roche’s Diagnostics Division.
C. Roche press release about the launch of uPath, incorporating the LVM technology ( https://diagnostics.roche.com/global/en/news-listing/2019/roche-launches-uPath-enterprise-software.html; 15th Jan 2019).
D. Letter about the LVM’s influence on investment, development and sales of a new scanner for digital pathology, from the Chief Executive Officer, FFEI Ltd (2nd Apr 2020).
E. Letter from Leeds Teaching Hospitals NHS Trust (LTHT) Pathology Clinical Director.
F. Letter about the LVM’s influence on investment in new display products, from Barco’s Strategic Product Manager (12th May 2017).
G. Letter about the LVM’s influence on investment in new digital pathology software products, from Global Product Manager, Digital Pathology, Sectra Imaging IT Solutions, Sweden (19th Feb 2020).
H. Press release from the UK Government Department for Business, Energy & Industrial Strategy: “Artificial Intelligence to help save lives at five new technology centres” ( https://www.gov.uk/government/news/artificial-intelligence-to-help-save-lives-at-five-new-technology-centres; 6th Nov 2018).
I. Grant announcement for the UKRI Centre for Doctoral Training in Artificial Intelligence for Medical Diagnosis and Care ( https://gtr.ukri.org/projects?ref=EP%2FS024336%2F1).
Additional contextual information
Grant funding
Grant number | Value of grant |
---|---|
NEAT L004 | £591,506 |
NIHR - New and emerging uses of technology L004 | £800,000 |