Our mission is to ensure implementation of AI in healthcare benefits patients. This is no easy task. It will require the close collaboration of multiple industry partners, universities and clinical departments. RAIT is the framework which allows for a smooth collaboration on current and future projects by standardizing the interactions.

We address the logistical, regulatory and workflow barriers to implementation, so that AI can benefit patients timely, consistently and safely.


Osteoarthritis of the knee (KOA) is a common disease, which causes pain, reduces range of motion and affects the quality of life. With more than 10% of the danish population affected by osteoarthritis, the disease is a significant economical burden on society, due to reduced ability to work, increased number of sick days and high cost of healthcare services.

Radiology in Denmark has seen an exponential rise in the number of exams and images per exam, due to the increased use of computed tomography and magnetic resonance imaging. The number of radiologists did not rise correspondingly. This has lead to a de-prioritization of images, like radiographs of the knee, which are deemed non-time critical. In fact, many of these images, especially at smaller hospitals are not read by radiologists at all.

In this project we investigate the clinical impact of an AI algorithm, which can automatically assess the degree of osteoarthritis of the knee. We want to test if the algorithm increases the quality of such radiograph readings . Also we are testing if an implementation of the algorithm can reduce the number of uneccessary referrals for an MRI scans of the knee.


Chest x-rays are the most commonly performed radiological imaging worldwide, with around 650.000 yearly examinations in Denmark. Moreover, there is an increasing usage of CT and MRI in Danish healthcare, which is often prioritized above conventional x-rays. Combined with a general shortage of radiologists, chest x-rays are often only reported on several days after radiography was performed.

Recent developments in artificial intelligence (AI) – in particular deep learning – has greatly enhanced computer aided detection in chest x-rays with performance on par with experienced radiologists in selected areas. In the SmartChest project, we investigate the potential of implementing such AI algorithms in Danish healthcare, starting with testing the accuracy of a validated algorithm from Enlitic Inc. (San Francisco, California) in Danish patients. This includes testing the ability to determine if the image is globally normal or abnormal. If abnormal, urgency of findings are assessed. Furthermore, we wish to evaluate whether output from the AI algorithm can augment the reading of chest x-rays by radiologists in training. We hypothesize, that relevant diagnostic accuracy can be achieved and that the algorithm therefore can increase the overall efficiency and quality of x-ray readings.