Table 2 Inclusion and exclusion criteria for selection of studies

Inclusion criteria

1. Included studies must cover domains/topics used—or mentioned as useful—for assessment of the value of AI in the area of medical imaging:

   a. Actual evaluation or assessment of the value of a specific AI technology, i.e. AI interpretation, classification or pattern recognition of an image

   b. Guidelines, statements, recommendations, white papers, evaluation models or (HTA) frameworks described in the literature and used for assessing the value of AI

   c. Reviews, surveys voicing future needs/research/challenges and future work when evaluating the value of AI

2. Studies must be published in English between January 1st 2016 and September 18th 2020

3. Setting: Articles written within a hospital setting or public healthcare organisations are included, e.g., hospitals, dentists, universities, etc

Exclusion criteria

4. Type of study or publication: All studies including grey literature, reports, and books but we exclude citations, patents, conference book, book of collected congress contributions, and conference abstracts

   a. Because of many hits at the full-text stage, we later excluded: opinion, commentary, or viewpoint articles

5. Studies only reporting on clinical efficacy or performance metrics, validation studies, technical development of the prediction model or AI-model, i.e., studies focusing on reporting on, sensitivity, specificity, diagnostic accuracy, precision, AUC, software used, the robustness of the model, etc

   a. Technical interpretation of images, including image optimisation and enhancing methods for highlighting images

6. AI on non-clinical images like images of surgical equipment, plants and animals or not on images, e.g. AI on electronic health records or workflows, AI in drug development, surgical robotics, electrocardiogram, thermographic scans, brachytherapy treatment, or augmented reality visualization and VR