Swedish AI system effective in diagnosing prostate cancer

Research validates solution backed by EIT Health

The Lancet Oncology has today published the results of a study demonstrating the accuracy of an artificial intelligence (AI) system in diagnosing prostate cancer in tissue samples.¹

STOCKHOLM 9 JAN 2020 The study, which was led by Karolinska Institutet in Sweden, showed that the AI system had considerable accuracy in determining whether a sample contained cancer, and in estimating the length of the cancer tumour in the biopsy. Furthermore, the AI system was comparable with 23 international, leading uropathologists in determining the Gleason score, the most important prognostic marker for prostate cancer.

“Our AI tool has the potential to reduce the workload of uropathologists, allowing them to focus on the most difficult cases and at the same time act as a safety net to improve quality.” said Martin Eklund, Associate Professor at the Department of Medical Epidemiology and Biostatistics at Karolinska Institutet. “It also has the potential to speed up diagnostics and reduce costs for healthcare services.”

The AI system has been developed by the same team that launched the new blood-based prostate cancer diagnostic test Stockholm3 in 2017. Stockholm3 doubles the number of aggressive cancers that are identified, whilst also reducing the number of unnecessary biopsies by 50% compared to current clinical practice with PSA.² Stockholm3 is currently used in clinical practice in Sweden, Norway, Finland and Denmark.

EIT Health has played a key role in accelerating the development and implementation of Stockholm3 and the new AI system, OncoWatch through network and financial support. As part of the development of OncoWatch, the AI system will be validated in 2020 in a nine-country multicentre study to assess its performance across different labs and in a wider range of digital pathology scanners. A first CE-marked product is expected to launch by the end of the year.

“There is a high demand for improved tools in prostate cancer diagnostics and we have proven that we can take new innovative tests to the market. The Stockholm3 test was used in more than 20,000 men in the Nordics alone in 2019, and I believe that our new AI system, OncoWatch, will also have strong uptake.” said Martin Steinberg, Project Leader for EIT Health-backed OncoWatch.

Prostate cancer is the most frequently diagnosed cancer among men in Europe, with approximately 450,000 new cases each year.³ Early diagnosis and treatment of aggressive prostate cancer is crucial to survival.

“Prostate cancer blights the lives of a large number of men in Europe each year, and despite the advent of new and innovative diagnostics and treatments, we still see too many men die from the condition.” comments Jan-Philipp Beck, CEO of EIT Health.

“Prostate cancer is often treatable with early diagnosis and effective treatment, and AI represents a large opportunity to complement existing medical expertise within the diagnostic process. The results of the study demonstrate the exciting developments that are happening in the field of AI, and how the technology can be applied to medical practice to improve the lives of patients and citizens in Europe. EIT Health is proud to support both the Stockholm3 and OncoWatch projects in the development of improved diagnostic tools to tackle prostate cancer.”

About OncoWatch

OncoWatch is an EIT Health-backed project that aims to provide game-changing improvements to prostate cancer diagnostics and treatment by combining big data, AI and cloud-based technologies to achieve several advances in the imaging of prostate biopsies, analytical tools for precision medicine to support effective treatment, and cloud-based infrastructure for easy access to the platform anytime, anywhere. The project is led by EIT Health partners the Karolinska Institutet and OncoAlgorithm in close collaboration with Philips, Synlab, Thermo Fisher and the European Prostate Cancer Coalition, a patient organisation.

References

1. The Lancet Oncology. Artificial intelligence for diagnosis and grading of prostate cancer in biopsies: a population-based, diagnostic study, January 2020.
2. Bergman M, et al. Män som vill testa sig för prostatacancer – en strukturerad modell Läkartidningen;115: FCDT, October 2018.
3. Ferlay, J. et al. Cancer incidence and mortality patterns in Europe: Estimates for 40 countries and 25 major cancers in 2018. European Journal of Cancer, November 2018.