Partners from the EuCanImage consortium met in early July 2022 at the University of Barcelona for a project-internal data annotation workshop to ensure that all requirements and functionalities are aligned with both the clinical and technical needs.
During the workshop, we recorded recorded interviews with project coordinator Karim Lekadir of the University of Barcelona, Maciej Bobowicz of the Medical University of Gdansk and Pär Kragsterman of Collective Minds Radiology.
The interviews highlight the importance of aligned data annotation and the main challenges in achieving good data annotation. The full video will be published shortly.
One of the main challenges that we would like to address in this project is the challenge of data annotation, specifically before the cancer imaging data can be provided by the hospitals to the researchers, they need to be extracted from the hospital records, they need to be anonymized, prepared and annotated, which means that radiologists in the hospital need to visualise and explore the images, identify where the tumours and the legions are and label them.
Karim Lekadir
Standardised data annotation provides the data sets that follow the same rules, so it’s much easier for the AI developers to harmonise that data that is provided in a standardised way according to a specific set of guidelines.
Maciej Bobowicz
We are a team of highly specialised individuals and teams from across Europe that are really one of the best in the world at what they do. We have AI expertise in the room, we have data scientists, medical doctors, radiologists, and surgeons in the room. We have software developers and companies around this that can together in a collaborative way build probably one of the most advanced solutions today.
Pär Kragsterman
