Although a lot of developments are made in the area of artificial intelligence (AI) in radiology and the number of publications is ever-increasing, the adoption in clinical practice still seems to lag behind and is limited to a small number of tools. To analyze this situation, we conducted a so-called technographic study. In this study, we identified 269 AI applications from 99 different companies. An interesting observation at the time of our study was the fact that all applications were “narrow-AI”, only capable of performing a single radiological task in one single image modality type. We also saw that, although already prompted as AI applications, 44% were either in the test or development phase and not available commercially. Of the available applications, 60% did not have any regulatory approval. Our analysis shows an increasing trend in the number of applications and an increasing percentage of applications cleared for clinical use.
Our technographic study clearly shows that there is great potential for AI in diagnostic radiology, but clinical implementation of AI in diagnostic radiology is dramatically lagging behind on the development of new applications that can offer extensive features and functionalities. The collaboration of the radiological community with companies is essential to help them develop the right tools and to ensure such quality that regulatory approval is warranted.
- Many AI applications are introduced to the radiology domain and their number and diversity grow very fast.
- Most of the AI applications are narrow in terms of modality, body part, and pathology.
- A lot of applications focus on supporting “perception” and “reasoning” tasks.
Authors: Mohammad Hosein Rezazade Mehrizi, Peter van Ooijen & Milou Homan