Background:
BreastScreen Norway invites all women aged 50-69 to mammography screening every two years with the aim of reducing breast cancer mortality. To achieve this purpose, it is important that the mammography images are of high quality, which is a responsibility entrusted to the radiographer capturing them. Today, all breast centers have one or more dedicated radiographers who regularly assess the images of all the radiographers at the breast centre. This assessment is retrospective, with a selection of images from each radiographer categorized as "perfect," "adequate," or "inadequate." Additionally, individual areas for improvement are identified. An assessment requires around 5-10 minutes per examination and subjective variations in assessments are expected both within and between breast centres.
It is now possible to use artificial intelligence (AI) to automate the quality assessment of mammography images, providing objective feedback on each individual image taken by the radiographer. This means that radiographers can receive more and more consistent feedback about the technical image quality from AI.
Objectives:
The project will investigate new methods for monitoring and improving image quality in BreastScreen Norway. An AI-based solution that assesses and provides feedback on technical image quality will be compared with a traditional training program consisting of lectures and assignments. Automatic quality assessments before, during, and after the intervention period will be compared to evaluate the effect of the various interventions (Figure 1). The results will form the basis for concluding and recommending the best solution for implementation after the project, which is the long-term goal.
Figure 1: Schematic representation of the study design with different interventions at the University Hospital of Northern Norway HF (UNN) and the Nordland Hospital Trust (NLSH). Collection of quality parameters will be done before, during, and after the intervention period to examine the effect in both groups.
Project partners:
This is a collaborative project between UNN, NLSH, the Norwegian Cancer Registry, Helse Nord IKT and UiT/Visual Intelligence.
