PhD positions in Device-Related Thrombus Modelling and Machine Learning in Left Atrial Appendage Occluder Device Optimisation

Two PhD positions are available at the BCN MedTech Research Unit (https://www.upf.edu/web/bcn-medtech/), Department of Information & Communication Technologies (DTIC) of the Universitat Pompeu Fabra (UPF), Barcelona, Spain, as part of the EU Horizon 2020 SimCardioTest project (2021-2024; SC1-DTH-06-2020, Grant agreement No. 101016496), ideally starting September 2021 (earlier start should also be possible). SimCardioTest is a recently granted European project involving a large consortium of clinical, industrial and academic partners (see Figure 1), with the main goal of developing a standardised and secure cloud-based platform where in-silico trials run seamlessly. UPF is the leader of the cardiac use case related to left atrial appendage occluder (LAAO) devices, where we will demonstrate the platform effectiveness, along with the required verification and validation processes and certification support to optimise LAAO device settings.

PhD project 1 (PhD1): Computational models for the prediction of device-related thrombus in LAAO devices

In collaboration with other partners of the consortium (mainly Boston Scientific, IHU Bordeaux and Simula), as well as several junior/senior researchers at UPF, PhD1 will work on the development of computational models to simulate the thrombus formation process after LAAO device implantation, using cascade and agent-based models, to couple with computational fluid dynamics (CFD) simulations. Fast fluid simulations with GPU-based solvers or deep-learning will also be investigated. A portfolio of realistic CAD models of LAAO devices will be created. Additionally, the effects of (anti-coagulants, anti-platelets) drug therapies will also be modelled. Virtual contrast agents and porous flow modelling through device frames after implantation will also be investigated to characterise acute response. PhD1 will also contribute to the generation of a large virtual population of in-silico fluid simulations and the development of quantitative metrics to characterise flow dynamics. These tasks will leverage on the past research by UPF members on LAAO-based fluid simulations (e.g., Aguado et al., 2019, https://doi.org/10.3389/fphys.2019.00237; Mill et al., 2020, https://doi:org/10.1016/j.cjca.2019.12.036) and agent-based models of atherosclerosis (Olivares et al., 2017, https://doi:org/10.1093/bioinformatics/btw551). PhD1 will be directly supervised by Pr Oscar Camara and Dr Jérôme Noailly from BCN MedTech at UPF.

PhD2 project (PhD2): Machine learning for the optimisation of device and drug therapy treatment in patients with atrial fibrillation

In collaboration with other partners of the consortium (mainly Inria, IHU Bordeaux, Boston Scientific), and junior/senior researchers at UPF, PhD2 will work on the development of machine learning tools to predict the optimal LAAO device settings for each individual patient. Morphological descriptors of the LA and LAA will be extracted to characterise their geometry, which will be combined with in-silico indices provided by fluid simulations run with multiple combinations of LAAO device settings. Unsupervised clustering methods such as the ones based on Multiple Kernel Learning, in which UPF has long-standing experience (e.g., Sanchez-Martinez et al., 2018, https://doi.org/10.1161/CIRCIMAGING.117.007138) will be explored to identify patterns among the data for a better understanding of the most relevant biomarkers predicting good prognosis. The use of deep learning techniques as surrogates of fluid simulations, based on our ongoing work (e.g., Morales et al., 2019, https://doi.org/10.1007/978-3-030-39074-7_17; Acebes et al., 2020, https://doi.org/10.1007/978-3-030-68107-4_4). PhD2 will be directly supervised by Pr Oscar Camara and Pr Gemma Piella from BCN MedTech at UPF.