OPEN Research Support

MD, PhD fellow
Jonas Asgaard Bojsen
Department of Radiology, Odense University Hospital

Projekt styring
Projekt status    Open
Data indsamlingsdatoer
Start 01.06.2021  
Slut 31.12.2023  

Accuracy validation of an AI for MRI stroke detection - a retrospective cohort study

Short summary

The purpose of the project is to determine the detection accuracy of stroke by the commercially available Artificial Intelligence (AI) system Apollo (Cerebriu A/S, Copenhagen, Denmark).


The workload of of magnetic resonance imaging (MRI) on radiology departments are steadily increasing each year, whilst the number of radiologists to plan, interpret and report the scans are the same. To aid the radiologists, AI systems have been proposed. Before any such system can be implemented, it must be thoroughly scientifically validated.

Description of the cohort

The cohort consists of up to 1000 patients 18 years or older, who have performed an MRI on Odense University Hospital (OUH) on the suspicion of stroke. MRI sequences FLAIR, DWI and either GRE T2* or SWI must have been performed, and the images must be of sufficient quality.

Data and biological material

The raw DICOM image sequences will be collected and pseudonymized before being handled by the AI software.

Processing of your personal data (danish):

Collaborating researchers and departments

Cerebriu A/S, Copenhagen, Denmark

    Department of Neurology, Odense University Hospital

    • David Gaist, professor, MD