The past decades have been characterized by fantastic progress in medical hardware and the innovations have enabled significant progress in radiologists’ clinical and diagnostic capabilities. The next wave of innovation with similar potential, at the very least, will, however, be software-based.
The tedious detection of cancer or the comparison of lesions in follow-up examinations in thousands of images will be assisted or wholly taken over by algorithms. The more AI algorithms are available, the larger the part of the workflow that can be automated, and the more quantitative data will be available to determine personalized treatment decisions.

Example of AI-assisted chest CT reporting view, enabling pre-filling of structured report elements based on AI-findings.
The common notion today is that “the future of radiology is Artificial Intelligence”. Why is it, then, that AI for image analysis has not yet found broad application in clinical routine? What are we waiting for? [1].
While the attention and expectations towards AI-based image analysis are certainly justified, there continues to be considerable challenges in making it a reality in clinical routine:
- Access to (structured) data for developing and training the algorithms
- Broadening the clinical scope of AI-based image analysis tools
- Integrating AI-based image analysis tools into the workflow
Establishing structured and fully integrated reporting will play an important role in overcoming these challenges. The creation of the diagnostic report is core to the considerable value that radiologists generate. However, current reporting continues to create unstructured texts. This lack of uniformity creates confusion for attending physicians and limits statistical analysis. It will also not support an unobtrusive integration of AI-based image analysis into the clinical workflow – which would be crucial, in light of ever-increasing pressure on productivity.
In Smart Reporting’s view, this non-standardized pattern represents a dead end in AI-based innovation. If the results of AI-based image analysis cannot be seamlessly integrated into the workflow, then the revolution will remain outside of the hospital gates.
That’s why at Smart Reporting, first and foremost, they start with the workflow of radiologists in mind. Used by thousands of radiologists around the world, their solution is a web-based platform for structured, fully machine-readable reporting [2]. Powered by medical templates from leading experts like Prof. Paul M. Parizel (Antwerp University Hospital) and Prof. Dr. med. Stefan Schoenberg (University Hospital Mannheim), it allows for reports to have up to 78% better completeness scores and that are up to 46% easier to read [3] than traditional reporting. Already today, this approach improves intra-hospital communication and confidence in clinical decisions by referring physicians.
Furthermore, Smart Reporting’s next-generation reporting engine will include Computer-Aided Detection (CAD) methods, which automatically analyze and interpret large sets of medical images. This results in an unprecedented level of seamlessly integrated reporting workflows.
Smart Reporting’s goal is to enable AI-assisted diagnostics in the future, and they believe the only way to get there is with the radiologist and the radiologist’s daily workflow in mind.
References:
- Wenn der Algorithmus dem Arzt sagt, dass der Patient eine Lungenembolie hat, Thomas Weikert, Neue Züricher Zeitung, 2019
- Smart Radiology (2019) Available at https://www.smart-radiology.com/. Last accessed 26 February 2019.
- Gassenmaier S, Armbruster M, Haasters F et al (2017) Structured reporting of MRI of the shoulder – improvement of report quality?, Eur Radiol 27: 4110–4119
Smart Reporting is part of the ECR 2019 Artificial Intelligence Exhibition (AIX) in hall X1, booth AI-13.
* Contents provided by Smart Reporting and adapted by the ESR office