Working at Medis – Internship

Deep Learning Internship

Explore deep learning approaches to pre-train a deep learning model for coronary angiograms segmentation and coronary bifurcations detection. 

Deep Learning Internship

Quantitative Flow Ratio (QFR) is a software for coronary angiograms analysis, employed for treatment and prognosis evaluation in stable coronary disease [1]. QFR employs a 3D reconstruction of an arterial segment based on two angiographic acquisitions, enabling accurate estimation of changes in coronary flow and detection of coronary obstructions. In the initial QFR releases, some manual input was necessary from interventional cardiologists to perform the analysis.  

Future versions of QFR will incorporate several AI techniques to eliminate the need for human interaction. However, due to the difficulty and time constraints in increasing our annotated data, we are exploring some alternative deep learning methods, that could offer a viable solution to fully exploit the potential of our dataset [2,3].  

Project Main Objective: Explore deep learning approaches to pre-train a deep learning model for coronary angiograms segmentation and coronary bifurcations detection. 


Skills to develop and/or put in practice:

• Programing skills in Python  
• Machine Learning: familiarity with Deep Learning frameworks (Pytorch, TensorFlow)  
• Medical Image processing and computer vision.   
• Academic writing. 
• Work in an interdisciplinary environment.  
• Critical and Analytical thinking.  
• Proactive attitude 

Work environment

You will be part of the Applied Research Department at Medis Medical Imaging, contributing with your own ideas and enthusiasm to enhance software that is already available for clinical use. Working with our teams offers you the opportunity to engage in scientific research within an international organization. Our team will provide support with their extensive knowledge in AI and algorithm development throughout the entire process. You will have access to GPU resources on our server and AWS for experiments. Additionally, you will be welcome to join us in the organization’s social events`  If you are interested and/or have any questions about this project, you can contact us by email.  


Person(s) of contact:

Balazs Borsos, MSc                                                              

Xikai Tang, PhD  

Jose Castillo Tovar, MD, PhD


1. Dobrić, M.; Furtula, M.; Tešić, M.; Timčić, S.; Borzanović, D.; Lazarević, N.; Lipovac, M.; Farkić, M.; Ilić, I.; Boljević, D.; et al. Current Status and Future Perspectives of Fractional Flow Reserve Derived from Invasive Coronary Angiography. Frontiers in Cardiovascular Medicine 2023, 10.     

2.     Krishnan, R., Rajpurkar, P., & Topol, E. J. (2022). Self-supervised learning in medicine and healthcare. Nature Biomedical Engineering, 6(12), 1346-1352.    

3.     Mathilde Caron, Ishan Misra, Julien Mairal, Priya Goyal, Piotr Bojanowski, and Armand Joulin. Unsupervised learning of visual features by contrasting cluster assignments. In H. Larochelle, M. Ranzato, R. Hadsell, M.F. Balcan, and H. Lin, editors, Advances in Neural Information Processing Systems, volume 33, pages 9912–9924. Curran Associates, Inc., 2020.

Can’t find your vacancy?