Commercial AI marketplaces for radiology

Start date: 30-11-2021
End date: 30-05-2022

Clinical Problem

The clinical implementation of Artificial Intelligence (AI) software in radiology is still in its early stages. The market for AI software in radiology is growing and there are over 180 CE-marked products commercially available [1]. These products target different diseases with different characteristics, such as modality, subspecialty and algorithms. This wide variety of commercially available AI products makes it increasingly complicated for radiologists and their departments to choose and contract between vendors to supply their different AI needs. Each department needs AI software with different characteristics and their own technical features and algorithms. Another challenge and extra administrative overhead is that the multiple AI software have to be integrated in existing clinical workflows and systems for different departments. This has led to a rise in marketplaces for AI software in radiology, which aim to offer multiple AI software on one platform. The software is implemented with the same workflow, which limits the required training for radiologists. Another advantage of AI marketplaces is that contracting invoicing may be concluded through one party. In the last years, the number of marketplaces for AI has expanded, with each marketplace distinguishing themselves by offering different features like platform and algorithm support, contracting and products [2]. This project aims to obtain a clear overview of all available marketplaces for AI in radiology that offer CE-marked products on one platform. The overview will be added on www.aiforradiology.com [1] and maintained to update the overview with the newest marketplaces. This overview aims to gain impact on a clinical aspect to make it easier and more transparent for radiologists to select the most fitting AI software for multiple departments. This will not only save selection and contracting time for the purchasers, but might also improve patient outcomes by providing best-in-practice AI software and products.

Solution

One of the main results of this project will be creating a new page on the website www.aiforradiology.com which lists a transparent overview of all available marketplaces that offer CEmarked AI software for radiology. Furthermore, the student will be involved in informing the Radboud UMC radiology department on the pros and cons of adopting a marketplace. If advising in favour of a marketplace, the student may advise on the best fitting marketplace for their needs. The thesis will be restructured as an article to be published. It is not mandatory for the student to assist in this process, but the option to assist is optional. Finally, another optional result of this project is a blog page on the website www.aiforradiology.com. This blog page should support users of the web page on how the overview has been established and how to operate the overview to find the best-fitting marketplace.

Data

Interviews will be held with different stakeholders in order to create the most transparent overview of marketplaces and important features. The stakeholders that will be included in this study are: - Marketplaces - Algorithm suppliers - Radiology departments - Clinical physicists The data collected from the stakeholders will be translated to a structured overview, differentiating between the various features. The data has to be verified by stakeholders and intelligible for all stakeholders. The analysis method is up to the student. However, the definition of marketplaces and included key features should be inquired and verified by the key stakeholders to reach a consensus regarding the overview.

People

Yassier Elawady

Yassier Elawady

Master Student

University of Twente

Kicky van Leeuwen

Kicky van Leeuwen

PhD Candidate

Diagnostic Image Analysis Group