In July 2023, Erasmus Medical Center’s radiology department kicked off a four-year project focusing on the use of CT to improve diagnosis and treatment outcomes of incidental pulmonary embolism (IPE). The project was made possible through a grant from the NHI (Netherlands Health Institute) and is a collaboration between contextflow and esteemed partners in the Netherlands:
- Erasmus Medical Center, Department of Radiology in Rotterdam
- Erasmus School of Health Policy and Management (ESHPM), Department of Health Technology Assessment in Rotterdam
- Technical University Delft, Department of Imaging Physics in Delft
The project is evolving under the guidance of the dedicated PhD student Erik Kempner at Erasmus University and will explore ways in which machine learning-based technology can be seamlessly blended into workflows to help radiologists and clinicians catch IPEs earlier and more accurately.
Besides an obvious benefit to patients and radiologists (and a seeming benefit to hospitals and the healthcare system at large) the use of AI in clinical practice has never really been explored for IPE. Therefore, the project will look into the impact of the technological advancement of IPE detection in various settings with the aim of redefining the landscape of radiology and potentially set new benchmarks in efficiency and patient-centered care as they relate to IPE.
This collaboration shows the power of interdisciplinary cooperation in medical technology innovation. Over the course of the next four years, the involved parties will work on producing results that could be used for setting new standards for radiology and healthcare.
What is IPE and why use AI to detect it?
Pulmonary embolism is a life-threatening condition that requires prompt diagnosis and treatment. Detecting IPE in computed tomography (CT) scans can be challenging, and the project group believes AI can help in improving the accuracy and consistency of IPE detection by analyzing and identifying subtle patterns that may go unnoticed by human observers and by reducing false negatives and false positives.
The interpretation of medical imaging can be time-consuming, especially in busy healthcare settings. AI algorithms can rapidly analyze images and assist radiologists in the detection of IPE, potentially reducing the time required for diagnosis.
By providing consistent and reproducible results, AI algorithms can also help reduce inter-observer variability and ensure that the same criteria are applied to every case. This standardization can lead to more reliable diagnoses, enabling better patient management and facilitating research and quality improvement initiatives.