Development of AI algorithms in the Mammography Programme

The Mammography Programme is undergoing research projects to develop artificial intelligence algorithms that can contribute to the assessment of mammography images.

Last updated: 9/21/2023

Background

Breast cancer is the most frequent cancer among women in Norway and in the world. Although some risk factors for the disease are known, the disease is difficult to prevent for the individual woman. Early detection through screening is therefore considered an effective tool in reducing mortality from the disease.

A standard screening examination in the Mammography Programme consists of mammography images from two angles of each breast. All mammography examinations are assessed by two X-ray doctors, both of whom indicate the images without knowing what the other has concluded (independent ambiguation). At the same time, very few participants in the screening programme have signs of breast cancer, and X-ray doctors therefore spend a lot of time assessing normal mammography images.

With recent advances in artificial intelligence (AI), there is potential to improve the current mammography programme, partly to support doctors in the assessment of images. By "artificial intelligence" we mean digital systems that learn to recognize patterns in mammography images that may be signs of breast cancer, because they have been trained to do so by analysing large amounts of data over time.

The Mammography Programme is undergoing two different research projects that use AI to develop new algorithms for assessing mammography images; "Machine learning in the Mammography Programme" (MIM study) and "AIforScreening; Robust and reliable AI for breast cancer screening" (AIforScreening). The Norwegian Computing Center is responsible for developing the algorithms.

Purpose

The overall purpose of the two projects is to develop AI algorithms that can be used as support for assessing images in the Mammography Programme.

The main objective of the MIM study is to develop a digital system that can pick out images that most likely show no signs of breast cancer. This can streamline and increase the quality of the Mammography Programme, among other things by allowing X-ray doctors to spend more time on women who have signs of breast cancer.

AIforScreening is a continuation of the MIM study and will further develop the AI algorithms from there to become even more robust and credible. The goal is that they can be used on mammography images taken with different types of imaging equipment, and that they should be able to use the information from previous mammography examinations in the assessment of relevant images.

Data basis

In order to develop good and accurate algorithms adapted to the Mammography Programme, large amounts of image data from the screening examinations are required. In addition, information related to attendance and findings on the mammography images is used, such as the X-ray doctors' assessments and the results of the screening examinations (screening information).

The MIM study is based on data from screening surveys conducted in the Mammography Programme at seven health trusts in the Northern Norway Regional Health Authority, the Central Norway Health Authority and the South-Eastern Norway Regional Health Authority. The surveys were conducted in the period from the introduction of digital mammography at the individual health trust (between 2004 and 2011) onwards.

In AIforScreening, the MIM data will be reused. In addition, image data and screening information will be collected from three new health authorities in the South-Eastern Norway Regional Health Authority.

The project group at the Cancer Registry of Norway will obtain image data from the collaborating health enterprises and screening information from the Cancer Registry's databases. The data will be pseudonymised and quality assured by the project group before handing over to the Norwegian Computing Center, which will develop and train the AI algorithms.

The project only uses data from women who have not opted out of having information from screening examinations with normal findings stored in the Cancer Registry, in accordance with the Cancer Registry Regulations. The women will not be contacted about the projects, and it will be impossible to recognize individuals in the results that are published.

Organization

The Cancer Registry of Norway is the project manager and data controller, and has the overall administrative responsibility for the projects. Among other things, we will ensure that the projects have all necessary approvals, collect image data and screening information in the project, test the AI algorithms and design any implementation plans in the Mammography Programme.

The Norwegian Computing Center is the data processor in the projects. They have a very high level of expertise in image analysis and AI, and are responsible for developing the AI algorithms.

The health trusts at the breast centres are partners in the projects. They are regional specialists in diagnosis and screening of breast cancer, and will assist with radiology expertise and knowledge of practical screening, as well as facilitate the extraction of imaging data.

The University of Tromsø has special expertise in IT systems for biological and medical applications, and will act as an important advisor on aspects related to this. They will also supervise master's students on related subjects.

Karolinska Institutet will contribute with its expertise in AI in mammography.

Status as of May 2023

In the MIM study , new algorithms have been developed as planned. The algorithms have also been tested on "new" datasets, i.e. a dataset that was not part of the data basis for the development work. Preliminary analyses suggest that AI algorithms have the potential to increase the sensitivity of mammography screening by increasing the number of cases detected by screening, reducing the number of interval cancer cases and reducing the workload of X-ray doctors.
The study has been completed by the Research Council of Norway and no longer has external funding. The project completion date in REC extended due to the need to further develop, validate and improve the algorithms. This work is ongoing.

In AIforScreening, necessary approvals are obtained and agreements signed. Since a large amount of new data (about 650,000 examinations with 2.6 million mammography images) will be added to the existing dataset from the MIM study, the collection is time-consuming. Extensive work is underway at the Cancer Registry of Norway to facilitate and implement this, in cooperation with external IT resources. Until the new dataset is ready, the Norwegian Computing Center uses the information from the MIM study to work with the project's objectives.