Vara collaborates with researchers at Karolinska Institutet for independent AI evaluation

Vara, the Berlin-based deep tech startup on a mission to provide every woman worldwide with life-saving access to better breast cancer screening, is today announcing a collaboration with researchers from Sweden’s world-renowned medical university Karolinska Institutet.

The objective of the collaboration is to independently evaluate Vara’s AI model for mammography screening, including comparisons with other similar AI systems. Following publication, the results will be used in the creation of a platform to validate AI systems being used in breast imaging, known as the VAI-B Platform (Validation of AI in Breast Imaging).

The VAI-B Platform is part of a Swedish-born project financed by Vinnova, Sweden’s innovation agency, and Regional Cancer Centers in Collaboration. It is intended to be used as a national and, possibly, international resource for the validation of AI systems in breast imaging. The aim is to create a scalable platform delivering value to hospitals: in terms of impact assessment and evaluations, and to AI developers, in regards external validation and benchmarking. The project emanates from a national incubator environment within the Analytic Imaging Diagnostics Arena (AIDA).

The project is led by Fredrik Strand, MD, PhD, and Sophia Zackrisson, MD, PhD. Dr Strand is a consultant radiologist within the Breast Imaging Unit at the Karolinska University Hospital, a docent at Karolinska Institutet, and is actively applying new machine learning techniques to breast radiology images. Prof Zackrisson is a senior consultant radiologist at Skåne University Hospital Malmö and a professor at Lund University, Sweden.

Stefan Bunk, Vara CTO, said: “We are excited to work with the team led by Dr Strand and Prof Zackrisson, both of whom are leaders in the field of breast imaging. The VAI-B project is an innovative European project we are eager to get started. Not only will it show how Vara performs independently in a real-world setting, but the project also sets a unique precedence in algorithm evaluation.”

The images and dataset in the pilot phase will come from three hospitals in different parts of Sweden, as well as from the academic work of Dr Strand and Prof Zackrisson. The aim is to have a diverse dataset representing different demographics and different mammography vendors.

Dr Fredrik Strand at KI said: “We intend the VAI-B platform to serve as a certification centre for AI algorithms in Sweden, and aim to build trust in AI applications used in healthcare. The platform does not have a one-off purpose, but can gather long-term data across Sweden to validate performance of new AI software versions. And importantly we ensure the entire process is GDPR compliant.“

About Vara: Better breast cancer screening should be a universal offering to every woman in the world. Vara’s AI-powered software platform, created with screening radiologists within the German Mammography Screening Programme, mitigates much of the human subjectivity associated with reading mammography results — and reduces the repetitive work screening physicians are routinely subjected to. Vara’s full-stack platform manages all processes within a screening centre and makes breast cancer screening more effective, more measurable, and more accessible for everyone, everywhere. By democratising access to early screening around the world, we deliver measurable impact backed by clinical evidence. We are partnering globally, to deliver breast cancer screening where it is needed most. For more information see www.vara.ai

About Karolinska Institutet and its Department of Oncology-Pathology: As a university, Karolinska Institutet is Sweden’s single largest centre of medical academic research and offers the country’s widest range of medical courses and programmes. The Department of Oncology-Pathology conducts basic, translational, clinical research and teaching activities related to cancer. Their aim is to discover fundamental mechanisms in all aspects of cancer biology, and find novel biomarkers and potent anti-cancer drugs for personalized cancer medicine. They maintain state-of-the-art equipment for scientists to share for more sustainable and efficient work. Read a brochure about the Department https://ki.se/media/144824/download and more about Dr. Strand’s research team https://ki.se/en/onkpat/research-team-fredrik-strand.

About the AIDA incubator: AIDA is a Swedish arena for AI innovation in medical image analysis, joining academia, healthcare, and industry. AIDA is an initiative within the Strategic innovation program Medtech4Health, jointly supported by VINNOVA, Formas and the Swedish Energy Agency. For more information on AIDA’s incubator branch, see: https://medtech4health.se/en/aida-arena/incubator/