DLMedIA: Deep transfer learning in cardiovascular disease

Developing deep transfer learning techniques to effectively analyze medical imaging data with variation in scanners, scan protocols, and patient populations.

ONGOING

2017

 

 | 

2022

 

RESEARCH AREAS

About project

Deep learning techniques have shown great promise in medical image analysis with an increasing number of reports showing that well-trained models outperform classical machine learning approaches in different medical applications. A drawback is still that deep models require a relatively large set of representative training data to perform well. In practice, truly representative training data is difficult to obtain and performance of a model will then deteriorate. The aim of this project is to develop deep transfer learning techniques to effectively analyze heterogeneous medical imaging data with variation in scanners, scan protocols, and patient populations. We focus on the analysis of cardiovascular CT images. The project is part of the TTW Perspectief DLMedia project with Philips Healtcare as our industrial partner.

 

Our publications & software