University of Twente Student Theses

Defining platforms for artificial intelligence algorithms in radiology : increasing the transparency of the market for AI platforms for radiology

Elawady, Y.M. (2022) Defining platforms for artificial intelligence algorithms in radiology : increasing the transparency of the market for AI platforms for radiology.

Full text not available from this repository.

Full Text Status:	Access to this publication is restricted
Embargo date:	31 August 2024
Abstract:	Introduction: Deploying multiple AI products for radiology is unmanageable in practice. The clinical added value of AI products in radiology is not achieved by deploying AI products as stand-alone applications. To tackle this problem, AI platforms are introduced to the market. However, due to its infancy, a definition of an AI platform remains absent. Combined with the unknown availability of AI platforms for radiology and their offered features, it is challenging for radiology departments to procure a fitting AI platform Goal The main goal of this thesis is to provide guide radiology departments in the procurement of a fitting AI platform. This goal is achieved by providing a definition for AI platforms and increase transparency on available AI platforms. Methods: A questionnaire is constructed and validated by a group of expert stakeholders using the DELPHI method. The questionnaire gathered data on features as offered AI products, deployment of AI products, output of AI products, workflow integration and support. An analysis of the AI platform features resulted in a definition of AI platforms to complement the literature. With the results, an online overview of all available AI platforms for radiology and their features is created on www.aiforradiology.com. Results: the overview included 35 AI platforms for radiology, offering the deployment of a wide selection of AI products. Out of the 10 responses from the questionnaire, all 10 AI platform vendors offer a solution for the processing of radiology images with third party AI products through the platform. All 10 platforms also offer an automatic orchestration of medical images to relevant AI products. Most platforms (8/10) offer the option for customers to develop and deploy AI algorithms on the platform and 9 out of 10 platforms offer the procurement of third party AI products through the platform. Discussion: an AI platform is defined as a technical solution that enables the deployment and orchestration of third party AI products, consolidating all service and support. One of the implications is that the group of experts to assess important features was not diverse enough. The thesis also did not focus enough on features that exclude AI platforms. It is recommended that similar research is conducted for AI platforms for pathology.
Item Type:	Essay (Master)
Clients:	Radboud University Medical Center, Nijmegen, The Netherlands
Faculty:	TNW: Science and Technology
Subject:	44 medicine
Programme:	Health Sciences MSc (66851)
Link to this item:	https://purl.utwente.nl/essays/92860
Export this item as:	BibTeX EndNote HTML Citation Reference Manager

Show download statistics for this publication

Repository Staff Only: item control page