Background:
The University Hospital of Northern Norway will carry out a pilot project to assess the use of artificial intelligence (AI) in diagnostic imaging. AI as decision support has in previous studies shown improved workflow in the department, shorter waiting times for patients, and reduced reporting time for radiologists. The specific needs outlined by the Northern Norway Regional Health Authority were detailed in the radiology report from November 2022. In this report, an interdisciplinary working group developed a priority list of disciplines where AI tools could relieve radiologists. The purpose of the pilot project is to gain practical experience with the introduction and use of AI solutions in clinical practice and ensure the quality and effectiveness of AI tools with local data. The project is scheduled to start in March 2024, with a duration of up to 6 months.
The applications are located on the Sectra Amplifier Platform, a cloud platform that works as a host and administrator for CE-marked third-party applications. A retrospective quality assurance of 3 AI algorithms is planned:
- Gleamer Bone View Trauma – Can detect fractures on X-rays of the skeleton
- Contextflow ADVANCE Chest CT – Can detect 19 disease patterns on CT of the Thorax
- Lunit INSIGHT Chest CT – Can detect 10 disease patterns on chest X-ray
Objectives for the test period are
- To gain practical experience with the introduction and use of AI solutions in clinical practice
- To ensure that the software works on local data
- Gaining experience with usability and workflow
- Increasing trust in AI
Project partners:
The pilot project is a collaboration between the X-ray department at UNN Tromsø, SPKI, HNIKT, and FSR. Validation will be done by radiologists at UNN Tromsø. Project leaders are AI radiographer Camilla H. Andersen and Karl Øyvind Mikalsen at SPKI. The project is anchored in the Northern Norway Regional Health Authority, the Strategy for Artificial Intelligence (2022-2025), and the follow-up report Artificial Intelligence in Radiology (2022).
