Open PhD positions

The BRIDGE-AI doctoral network is hiring 15 ambitious doctoral candidates at various locations in Europe (Finland, the Netherlands, Belgium, Germany, Switzerland and Austria). The aim of the network is to develop new and trustworthy AI methods for longitudinal neuroimage analysis, helping clinical experts assess temporal changes in patients with chronic brain disorders. More information about the project is available from the BRIDGE-AI website.

Apply now!

We offer:

  • A unique and stimulating environment to work on cutting-edge AI method development for computational brain imaging, with ample opportunities for collaborating closely with expert clinicians and world-leading companies in the medical imaging domain. You will be co-supervised by a multidisciplinary team, receiving valuable input from industry, academia, and hospitals alike.
  • A dedicated PhD training program with high-profile invited lecturers, combining both science-based and transferable skills training, with attention to your future career development. All BRIDGE-AI’s doctoral candidates will convene on regular intervals for training events organized in Helsinki (FI), Leuven (BE), Lausanne (CH), Bern (CH), Vienna (AT) and Eindhoven (NL).
  • Access to large collections of clinical neuroimaging data and expert guidance (including through extensive international research visits) for evaluating the true clinical relevance and impact of the AI tools you will be developing.
  • Full-time employment on the BRIDGE-AI project with competitive remuneration for three years. Depending on the host organization, subsequent funding needed to finalize your PhD thesis may be available.

Your profile:

  • A MSc degree in physics, computer science, electrical engineering or a similar degree. Candidates who expect to obtain their MSc degree in the near future are also welcome to apply.
  • A strong interest in both new AI method development and the real-world clinical context in which AI methods need to operate. Able to work in an interdisciplinary team and interested in collaborating with clinical partners.
  • A high level of perseverance.
  • Strong programming skills in e.g., Python/NumPy, PyTorch or C++.
  • Excellent communication skills in English (both oral and written).
  • An eagerness to travel within Europe.

Other prerequisites:

  • Applicants can be of any nationality, but must not have resided or carried out their main activity (work, studies, etc.) in the country of employment for more than 12 months in the 36 months immediately before their date of recruitment. Compulsory national service, short stays such as holidays, and time spent as part of a procedure for obtaining refugee status under the Geneva Convention are not taken into account.
  • At the date of the recruitment, applicants can not already be in possession of a doctoral degree. Researchers who have successfully defended their doctoral thesis but who have not yet formally been awarded the doctoral degree are also not eligible.
  • Successful applicants will need to enroll in a doctoral program at the PhD awarding institution associated with each position. Therefore, eligibility requirements (e.g., relevant MSc degree, language proficiency, …) for the relevant institution need to be fulfilled. Please see the FAQ page for institution-specific requirements.

To apply:

  • Applications can only be submitted through Aalto University’s online job platform until 30 January 2026 at 23:59 Finnish time (GMT +2), or until all positions have been filled (whichever comes first).
  • Screening and filling of the positions will start as soon as applications are received. Therefore, it is recommended to apply as early as possible.
  • We strive to ensure diversity and gender equality in the BRIDGE-AI network through an open, transparent, and merit-based recruitment. Women and others underrepresented in the field of computational neuroimaging are particularly encouraged to apply.
  • In order to apply, you should include: (1) a cover letter explaining your motivation for applying; (2) your CV; (3) relevant transcripts of studies and certificates of your degrees; and (4) the names and contact information of at least two professional references who may be contacted regarding your application. You should indicate which position(s) you are applying to (maximum 3), and indicate the order of preference within the selected positions (1 = highest preference).
  • We reserve the right to leave positions open, to extend the application period, and to reopen the application process.
  • For more information about the application and selection procedure, please see the FAQ page.

Employer: German Center for Neurodegenerative Diseases (DZNE)  

PhD enrollment: University Bonn 

Main supervisor: Prof. Dr. Martin Reuter 

PhD project description:  

Are you passionate about advancing neuroscience and clinical applications through cutting-edge AI technology? Do you want to contribute to tools that can transform how we understand brain function, study neurological diseases, and evaluate large-scale population studies? Join us to develop innovative AI methods for cortical surface modeling – an essential foundation for modern neuroimaging research. 

Why This Matters: The cerebral cortex, with its intricate folds and complex structure, holds key insights into brain health and disease. Accurate modeling and analysis of the cortical surface enable breakthroughs in understanding brain development, detecting abnormalities, and tracking disease progression. Your work will help build powerful AI tools that can improve diagnosis, guide treatment strategies, and push the boundaries of neuroscience research. 

In this PhD project, you will: 

  • Develop advanced AI methods to segment brain cortical atlases based on surface data, moving beyond traditional approaches. 
  • Create algorithms to detect anomalies on the cortical surface, such as Focal Cortical Dysplasia (FCD), aiding early diagnosis of neurological conditions. 
  • Innovate deep learning solutions for longitudinal surface extraction, tracking brain changes over time with unprecedented accuracy. 
  • Implement state-of-the-art deep learning techniques for cross-subject and cross-time surface registration. 
  • Thoroughly validate your methods to ensure broad applicability in neuroimaging workflows. 

Your contributions will elevate the accuracy, reliability, and generalizability of cortical surface analysis, setting new standards for the field. They will generate impact through their possible integration into widely used neuroimaging pipelines such as FastSurfer and FreeSurfer.  

What We Offer: We are the developers of FastSurfer and co-developers of FreeSurfer, world-leading open-source neuroimage analysis pipelines. We are experts in developing advanced deep-learning methods for medical image processing. At our lab you will have access to excellent supervision and to a powerful computational environment, including a high-performance cluster with close to one hundred NVIDIA 32GB GPUs and four dedicated servers reserved exclusively for our group, each with 8 GPUs (32-40GB RAM each). Furthermore, we offer research stays at world-renowned institutions, including the Martinos Center at MGH and Harvard Medical School, Boston USA, as well as Icometrix, Belgium. We provide a dynamic, collaborative research environment with mentorship from experts in AI, neuroimaging, and clinical neuroscience. 

Who You Are: 

  • You hold a strong background in computer science, biomedical engineering, applied mathematics, or a related field. 
  • You have experience in machine learning and deep learning with applications in medical imaging or neuroscience. 
  • You are motivated to work on impactful research that bridges AI development with real-world neurological applications. 
  • You possess solid programming skills (Python and PyTorch) and optimally have worked with geometric representations, such as triangle meshes. 
  • Experience with brain MRI data, or longitudinal analysis is a plus but not mandatory. 

If you’re eager to push the frontiers of AI in neuroscience and contribute to transformative research with world-class collaborators and resources, we want to hear from you! 

Foreseen secondments: For this PhD project, research visits are foreseen to:  

  • Dr. Dirk Smeets (1 month) at Icometrix (Leuven, Belgium) – industry leader on advanced methods for medical image analysis. 
Apply now!

Employer: icometrix, Leuven, Belgium  

PhD enrollment: UC Louvain, Brussels, Belgium  

Main supervisor: Dr. Diana Sima 

PhD project description:  

A central challenge in neuroimaging is the accurate tracking of subtle, longitudinal changes in the brain to monitor health and disease progression. While current computational models have shown success in analyzing healthy-looking adult brains by considering all time points simultaneously, a significant gap remains in applying these techniques to clinical scenarios. The presence of large abnormalities and lesions, such as those resulting from brain tumours, multiple sclerosis, or resection cavities, is currently an unsolved problem for existing methods. These pathologies can confound standard analysis pipelines, making it difficult to disentangle genuine anatomical changes, like brain atrophy, from the evolution of the lesion itself.  

This PhD project aims at developing a novel, disease-agnostic volumetric segmentation framework specifically designed for analyzing longitudinal scans in the presence of large abnormalities. The project will focus on developing robust deep-learning registration techniques for multimodal MRI, creating generative models to synthesize longitudinal lesion evolution patterns in multimodal MRI to generate training data, and developing models that simultaneously monitor both lesion load and brain atrophy. The expected outcome is a set of powerful tools capable of distinguishing genuine disease progression from stable conditions, thereby providing reliable measurements of brain atrophy and lesion evolution across various neurological disorders. 

Foreseen secondments: For this PhD project, research visits are foreseen to:   

Apply now!

Employer: University of Lausanne, Lausanne, Switzerland 

PhD enrollment: Life Sciences Program, University of Lausanne, Lausanne, Switzerland 

Main supervisor: Prof. Meritxell Bach Cuadra 

PhD project description: Paediatric low-grade gliomas are the most common brain tumours in children, representing up to 50% of all paediatric tumours of central nervous system. Although survival rates are excellent, these tumours are the leading cause of long-term neurological and cognitive morbidity due to their location and the intensive multimodal treatments (surgery, chemotherapy, radiotherapy) required. Children often need lifelong MRI follow-up, yet current evaluation mainly relies on manual lesion size measurements, offering limited sensitivity to subtle or diffuse brain changes. 

This PhD project aims to revolutionise follow-up imaging in pLGG through advanced AI-based quantitative analysis of longitudinal MRI data. The candidate will develop automated methods for multi-parametric change detection, assessing both tumour evolution and non-lesional brain alterations (e.g., atrophy, abnormal voxel-wise patterns). A central concept is the creation of a personalised “healthy digital twin” of each patient’s brain—an AI model enabling precise comparison of patient changes over time. To do so, the project will explore deep learning techniques for 3D tumour segmentation, domain-shift robustness, and uncertainty quantification to enhance clinical reliability. 

 The expected outcomes include: 

  • An open-source tool for quantitative change detection across multiple MRI contrasts. 
  • A framework to generate uncertainty maps from voxel to patient level, supporting safer clinical interpretation. 

By integrating clinical insight with cutting-edge AI, this project will contribute to advance assessment of response criteria in complement to current Response Assessment in Pediatric Neuro-Oncology (RAPNO) recommendations, with accurate volumetric measures with confidence intervals and individualized monitoring and ultimately improve long-term outcomes for children with brain tumours. 

What we offer 

  • A multidisciplinary project jointly with Lausanne University Hospital between cutting-edge brain imaging and advanced image processing, machine learning, and clinical applications. 
  • A dynamic, interdisciplinary, and international team of very motivated people.  
  • A stimulating work environment.  
  • Access to cutting-edge technology and state-of-the-art resources. 

Who You Are  

  • A master’s (MSc) degree in physics, computer science, or electrical engineering, or similar degree with an equivalent academic level. 
  • You have experience in machine learning and deep learning with applications in medical imaging.  
  • A strong will to develop clinically actionable methods and to interact with clinicians is required. 
  • Good programming skills Python, including full stack and deep learning frameworks (PyTorch or TensorFlow). 
  • Experience with brain MRI data, or longitudinal analysis is a plus but not mandatory. 
  • Good skills in English (oral and written) are required. Knowledge in French is a plus. 

Foreseen secondments: For this PhD project, research visits are foreseen to: 

  • Dr. Jonas Richiardi (3 months) at the Radiology Department (CH) 
  • Prof. Koen Van Leemput, (3 months) at Aalto University (Finland) 
  • Prof Roland Wiest (3 months) at University Hospital Bern (CH) 
  • Dr Bénédicte Maréchal (2 months), Siemens Healtheneers (CH)
Apply now!

Employer: icometrix, Leuven, Belgium  

PhD enrollment: Aalto University, Espoo, Finland  

Main supervisor: Dr. Diana Sima  

PhD project description:  Effective clinical management of Multiple Sclerosis (MS) relies on longitudinal Magnetic Resonance Imaging (MRI) to track disease progression through measures of brain atrophy and white matter lesion evolution. However, the diagnostic precision of this approach is significantly undermined by technical variability between MR scanners and a lack of data harmonization. This challenge is exacerbated in real-world clinical settings by the heterogeneity of acquired images—varying in contrasts, voxel resolutions, and scanner-specific properties—often occurring within a single patient’s longitudinal data. These inconsistencies introduce scanner-related biases that can obscure or mimic genuine pathological changes, creating a critical and unresolved barrier to reliable, long-term patient monitoring. 

This PhD project will directly address the problem of data heterogeneity by developing novel deep learning (DL) methodologies for the harmonization and analysis of longitudinal MRI data in MS. The primary objective is to create image-level harmonization techniques that enable robust, reliable tracking of annualized brain volume loss and changes in white matter lesion volume. To achieve this, the research will investigate promising DL-based approaches, such as adversarial learning, for targeted scanner-bias reduction. A key component of this work will be the creation of a comprehensive validation framework using real-world clinical brain images to benchmark MS lesion evolution estimation. The expected outcomes are robust modules for both brain atrophy computation and lesion evolution analysis, designed for implementation within the longitudinal icobrain ms software, while also advancing knowledge on uncertainty estimation and the explainability of DL models in clinical applications. 

Foreseen secondments: For this PhD project, research visits are foreseen to:   

Apply now!

Employer: Aalto University, Espoo, Finland

PhD enrollment: Aalto University, Espoo, Finland

Main supervisor: prof. Koen Van Leemput

PhD project description: Compared to clinical 1.5T or 3T MRI systems, ultra-high field MRI has a much higher contrast-to-noise ratio, enabling imaging at a much higher spatial resolution to depict fine anatomical details that cannot be visualized at lower field strengths. Despite the clear clinical value of 7T, there currently is a dearth of computational tools that can robustly quantify brain morphology at 7T. This is both because 7T scans contain specific imaging artifacts not seen in conventional MRI (e.g., strong intensity inhomogeneities and even local signal loss), and because of the lack of standardization in acquisition protocols across 7T sites.

The overall aim of this PhD project is to develop accurate longitudinal segmentation tools that work out-of-the-box on 7T MRI data acquired across different imaging sites, and to evaluate their clinical impact compared to currently available tools working at 3T. Towards this goal, you will develop novel generative models for brain MRI segmentation at 7T; compute automatic error bars for each segmented structure by designing efficient Markov chain Monte Carlo samplers; and use the computed error bars to decide which input contrast or combinations (MP2RAGE, MPRAGE, T2w, FLAIR, SWI) is better for atrophy estimation at 7T.

The project should result in an open-source software tool that can be included in the widely-used computational neuroimage analysis package FreeSurfer.

Your profile: This position requires strong skills in both math (e.g., linear algebra, Bayesian statistics) and coding (e.g., Python/NumPy, C++). Prior experience with medical imaging is not required.

Foreseen secondments: For this PhD project, research visits are foreseen to:

Apply now!

Employer: Aalto University, Espoo, Finland

PhD enrollment: Aalto University, Espoo, Finland

Main supervisor: prof. Koen Van Leemput

PhD project description: There is a great unmet need for automated tools to quantify differences between normal and abnormal early development in infant brain MRI, because aberrant morphological measurements have been associated with various neuropsychiatric, neurological, and developmental disorders. At present, however, the capabilities of state-of-the-art methods for automatic brain quantification in infants are significantly lagging behind those available for adult brain MRI. This is because analyzing infant brain MRI is considerably more challenging, due to the diminutive scale of the infant brain relative to the voxel resolution, the dramatic dynamical changes in neuroanatomy and MRI contrast as the brain matures, and the large diversity of image resolutions and scanner/sequence-specific contrast properties in data acquired at different institutions.

The goal of this PhD project is to significantly advance the state in the art in infant brain MRI segmentation, and to start applying the developed methods and tools to large clinical datasets of both normal and abnormal brain development. Towards this goal, a new generation of generative probabilistic models will need to be developed that can be efficiently inverted to obtain robust infant brain segmentations across a wide age range (0-5 years) and across a diversity of scanners and imaging protocols. The project should result in an open-source software tool that can be included in the widely-used computational neuroimage analysis package FreeSurfer.

Your profile: This position requires strong skills in both math (e.g., linear algebra, Bayesian statistics) and coding (e.g., Python/NumPy, C++). Prior experience with medical imaging is not required.

Foreseen secondments: For this PhD project, research visits are foreseen to:

Apply now!

Employer: icometrix, Leuven, Belgium  

PhD enrollment: Aalto University, Espoo, Finland  

Main supervisor: Dr. Simon Van Eyndhoven 

PhD project description: A significant challenge in large-scale neuroimaging studies is the robust segmentation and analysis of images acquired from different centers, each with unique imaging protocols and scanner platforms. This variability introduces domain shifts that can degrade the performance of machine learning models, limiting their generalizability. To overcome this, methods that can learn representations of brain MRI scans that are invariant to scanner-specific characteristics but retain crucial anatomical and pathological information are needed. Self-supervised learning (SSL) presents a promising avenue to achieve this by pre-training models on vast, unlabelled datasets to develop foundational representations that can be effectively adapted to a wide range of subsequent, or downstream clinical prediction tasks across different diseases. 

This PhD project will focus on developing novel self-supervised learning methods to create generalizable representations from heterogeneous brain MRI scans. The primary objective is to leverage SSL by pre-training a model on thousands of diverse MRI scans, enabling it to serve as a powerful auxiliary for various downstream tasks such as brain age prediction, lesion segmentation, and the prognosis of disease evolution. A key component of this project will be a rigorous evaluation of different SSL techniques and pretext tasks to establish their benefits and formulate standardization guidelines for their application. The expected results include a robust, pre-trained model ready for application in diverse clinical tasks and a comprehensive set of guidelines for applying SSL in medical imaging. The utility of this model will be demonstrated through a secondment at DZNE, where it will be applied to an in-house dataset to showcase improved training efficiency and performance against existing benchmarks. 

Foreseen secondments: For this PhD project, research visits are foreseen to:   

Apply now!

Employer: Medical University of Vienna

PhD enrollment: Medical University of Vienna

Main supervisor: Prof. Dr. Georg Langs

PhD project description: Are you interested in linking medical imaging data, genomics data to predict the progression and possible treatment response of brain tumor patients? Do you want to perform cutting-edge basic research in how we can use machine learning, multi-agent systems, and multi-modal data capturing molecular mechanics, and rich medical imaging to discover mechanisms, and to predict individual patient trajectories. Join our team to develop new machine learning methodology at the interface of medicine and tumor biology. 

Why this matters: The improvement of care for patients with brain tumors is a pressing topic, and we need to understand the link between what we can observe in patients, and the underlying biology of the disease to effectively steer treatment, and to create opportunities for identifying novel treatment targets. 

In this PhD project, you will: 

  • Develop new machine learning methods to analyse brain tumor data including genomics-, imaging-, and clinical data.
  • Develop novel approaches in the area of generative AI and multi-agent systems to identify mechanisms, and to develop prediction models for disease progression, and treatment response in individual patients.
  • Collaborate with machine learning researchers, imaging- and pathology-, and cancer experts.

What we offer: 

  • Co-supervision of your PhD by leading experts: Supervision of the PhD by Prof. Georg Langs, co-supervision by Prof. Adelheid Wöhrer (Univ. Innsbruck) and expert in molecular pathology, and Dr. Johannes Hofmanninger, leading machine learning research as the head of Research and Development at contextflow GmbH
  • You will be a member of the Computational Imaging Research Lab at Medical University of Vienna, and have an appointment at the Comprehensive Center for Artificial Intelligence in Medicine, an interdisciplinary research center with over 120 scientists working in the area of AI and medicine.  
  • You will be embedded in an interdisciplinary research team with members from machine learning, biomedical technology, medical imaging, and cancer medicine. Your interactions will include close ties to clinicians for feedback, collaboration, and the development of ideas. Students from these fields will be in the same lab.   
  • Cutting-edge computing environment including large-scale HPC/GPU cluster for deep learning

Who you are: 

  • Master in Computer Science, Machine Learning or related fields
  • Experience and interest in any of medical image analysis, deep learning, or bioinformatics
  • Motivated to develop novel machine learning approaches in the field of medical imaging and precision medicine. 

Foreseen secondments:: 

  • Prof. Jonas Richiardi, CHUV, University of Lausanne, (3 months), to gain experience in brain tumours, and computational modelling
  • Prof. Koen Van Leemput,  University of Aalto, (3 months), to receive technical supervision in medical image analysis (representation learning, generative modelling)
  • Dr. Johannes Hofmanninger, contextflow GmbH (1 month), to get familiarized with the commercial process from machine learning research to production software, learning about imaging- and non-imaging integration across domains.
Apply now!

Employer: Aalto University, Espoo, Finland

PhD enrollment: Aalto University, Espoo, Finland

Main supervisor: prof. Koen Van Leemput

PhD project description: The assessment of whether a fetus is at the expected development stage is currently performed in a rather primitive fashion in the clinic: clinicians manually perform 2D measurements from fetal brain MRI scans, and compare those to normative growth curves to see whether development is delayed. In this PhD project, you will significantly advance this state of affairs by developing new probabilistic generative models of the developing neuroanatomy across gestational ages, and use those models to accurately estimate the gestational age from fetal MRI scans.

This is a highly relevant but also very ambitious project: In order to be fully clinically useful, the methods and tools that you will develop should work “out-of-the-box” on data that is acquired with different scanners, image resolutions, reconstruction algorithms and acquisition protocols. Furthermore, your models should encode the cause-effect relationship between gestational age and the acquired MRI scans, so that the obtained predictions are inherently interpretable for clinicians (for instance, by generating counterfactuals simulating a particular fetus’s neuroanatomy at different gestational ages). Once you are confident in the performance of your methods and tools, you should also apply them to large clinical datasets of both normal and abnormal brain development.

Your profile: This position requires strong skills in both math (e.g., linear algebra, Bayesian statistics) and coding (e.g., Python/NumPy, C++). Prior experience with medical imaging is not required.

Foreseen secondments: For this PhD project, research visits are foreseen to:

Apply now!

Employer: Eindhoven University of Technology, Eindhoven, The Netherlands 

PhD enrollment: Eindhoven University of Technology, Department of Biomedical Engineering, Eindhoven, The Netherlands 

Main supervisor: Josien Pluim, Medical Image Analysis Group 

Clinical problem: An intracranial aneurysm is a bulging of a blood vessel in the brain. It carries a risk of rupture of the vessel, leading to a bleeding and possibly death or serious disability. It is estimated that 3% of the general population has an intracranial aneurysm. Not all aneurysms will rupture, however. Aneurysms can be treated by image-guided intervention, but that carries a significant risk of complications. Therefore, the aneurysm rupture risk has to be weighed against the treatment complication risk. Currently it is difficult to predict rupture, which limits clinical decision making and patient counseling.  Therefore, we urgently need better means to detect, predict and treat aneurysm instability for each patient individually.  

Your project: You will develop and employ AI methods to detect changes in aneurysms over time, based on Magnetic Resonance or Computed Tomography images of the patient’s brain. This will support so-called ‘watchful waiting’ in which a patient is regularly monitored and treatment is withheld unless rupture risk increases. You will design methods to characterise aneurysms from clinical images and combine those features with clinical parameters (e.g. age, history of smoking) to develop a rupture risk prediction model. Large language models may be used to automatically extract relevant information from free-text patients reports. You will ensure easy and reliable interpretation of the results for the interventional radiologists working with your models. The methods will be compared with the commonly applied PHASES score on a large clinical cohort of 1500 patients. 

You will regularly visit the University Medical Center Utrecht to meet in person with your clinical supervisor (prof. Irene van der Schaaf, interventional radiologist) and learn about the clinical aspects of your project. 

NB, a technical PhD project in the Netherlands spans four years and the appointment will be for four years. 

Foreseen secondments: For this PhD project, research visits are foreseen to:  

Apply now!

Employer: Medical University of Vienna, Spitalgasse 23, 1090 Vienna, Austria

PhD enrollment: Medical University of Vienna, www.meduniwien.ac.at 

Main Supervisor: Dr.techn. Roxane Licandro

PhD project description: Are you interested in trajectory learning of fetal brain development for early detection of brain malformation processes? Do you want to perform cutting-edge basic research in how we can use machine learning to discover mechanisms and morphometric patterns to predict individual brain development trajectories? Join our team to develop new machine learning methodology at the interface of computational modelling, fetal imaging, prenatal monitoring and neuroimage analysis. 

Why this matters: 

  • The potential discovery of new interpretable surface-based markers for brain malformation enables earlier detection of brain malformation trajectories and increases trust in AI for fetal image analysis.
  • The analysis of deviation patterns and attention maps provide insights into which regions are observed by the approach to classify a specific brain malformation leading to early diagnosis and potential treatment guidance.
  • This project aims at providing individual brain malformation classification scores including risk prediction of brain malformation emergence, suitable for the integration in clinical routine of prenatal diagnosis. 

In this PhD project, you will: 

  1. Create ML techniques to understand the trajectory of normative brain surface development based on retrospective in-utero MRI data of fetuses between gestation week 18 until term. Novel time-conditioned manifold learning techniques are developed to encode normative development of brain shape and cortical folding patterns. 
  2. Analyze novel methods to detect new shape and early imaging markers of brain malformation emergence by the analysis of morphologically deviating brain regions from the normative trajectory learned.  
  3. Compute a predictive score for brain malformation emergence risk using explainable AI techniques to identify the regions of the brain’s shape that trigger correct classification or emergence risk estimates of diverse brain malformation categories.

What we offer: 

  • Co-supervision of your PhD by leading experts: Supervision of the PhD by Dr.techn. Roxane Licandro, co-supervision by Prof. Gregor Kasprian (Medical University of Vienna) and expert in neuroradiology and fetal imaging, Prof. Dr. Georg Langs (Medical University of Vienna), and Dr. Tom Hilbert (Siemens Healthineers).
  • You will be a member of the Computational Imaging Research Lab at Medical University of Vienna, and have an appointment at the Comprehensive Center for Artificial Intelligence in Medicine, an interdisciplinary research center with over 120 scientists working in the area of AI and medicine.  
  • You will be embedded in an interdisciplinary research team with members from machine learning, biomedical technology, medical imaging, neuroradiology and fetal medicine. Your interactions will include close ties to clinicians for feedback, collaboration, and the development of ideas. Students from these fields will be in the same lab.   

Requirements

  • Master in biomedical engineering, medical informatics, applied mathematics, or any related subjects.
  • Experience with machine learning (esp. Python, MATLAB, R) and statistics.
  • Experience with medical imaging techniques (acquisition and processing).

Advantage 

  • Organization skills (scheduling, data management) 
  • Experience with shape-based analysis, GCNN
  • Previous experience in perinatal image analysis research
  • Affinity for teamplay

Foreseen secondments: For this PhD project, research visits are foreseen to:  

  • Dr. Meritxell Bach-Cuadra (UNIL) head section of the Lausanne University Hospital (CHUV) – University of Lausanne (UNIL) Signal Processing module of the Center for Biomedical Imaging (CIBM), named Trustworthy Medical Image Analysis Section. The aim of this secondment is to test brain malformation classification on external data and broaden the skills on manifold atlas learning strategies for fetal brain development trajectory encoding.
  • Dr Tom Hilbert (Quantitative Imaging Enthusiast & Acquisition and Reconstruction Expert at Siemens Healthineers, Switzerland), to get insights into MR imaging techniques, machine setups tailored to developmental cohorts.
  • Dr. Lilla Zollei, Associate Professor of Radiology (Massachusetts General Hospital, Harvard Medical School) to test brain malformation classification algorithms on external data with extension to perinatal postmortem brain image analysis.
Apply now!

Employer: Medical University of Vienna

PhD enrollment: Medical University of Vienna

Main supervisor: Prof. Dr. Georg Langs

PhD project description: Are you interested in understanding the capacity for reorganization in the human brain, and how it links to evolution, cognition, and disease? Would you like to investigate the evolution of cortical architecture, cognitive capabilities, and what it can tell us about how the brain changes, and compensates for effects of disease such as brain tumors? Join our team to investigate anatomical and functional brain imaging data of cancer patients, to study reorganization patterns of brain architecture, and investigate what brain evolution can tell us about the emergence of capabilities such as language, and the capacity of associated brain networks to reorganize. Experiment with embedding approaches that disentangle anatomical topography from the architecture of functional interaction patterns on the cortex, and with computational approaches that enable us to understand the origins and formation of these cortical structures. 

Why this matters: Our understanding of brain reorganization is instrumental for the design of novel therapies that support effective reorganization such as the recruitment of brain areas for language processing, if language centers are affected by tumors.  improvement of care for patients with brain tumors is a pressing topic, and we need to understand the link between what we can observe in patients, and the underlying biology of the disease to effectively steer treatment, and to create opportunities for identifying novel treatment targets. 

In this PhD project, you will: 

  • Develop new machine learning methods to analyse, track and model functional network reorganization paths on the cortex in brain tumor patients. 
  • Investigate the evolution of the cerebral cortex, and what it can teach us about the reorganization capacity of certain brain areas, with a focus on language processing as a test case. 
  • Develop a map of the cortex that identifies regions able to reorganize, landing zones for specific functional roles, and possible mechanisms driving this reorganization, and the associated cognitive outcome of patients 
  • Collaborate with machine learning researchers, imaging- and pathology-, and cancer experts.

What we offer: 

  • Co-supervision of your PhD by leading experts: Supervision of the PhD by Prof. Georg Langs, co-supervision by Dr. Meritxell Bach Cuadra, (Univ. Lausanne) an expert in neuroimaging and network analysis.
  • You will be a member of the Computational Imaging Research Lab at Medical University of Vienna, and have an appointment at the Comprehensive Center for Artificial Intelligence in Medicine, an interdisciplinary research center with over 120 scientists working in the area of AI and medicine.  
  • You will be embedded in an interdisciplinary research team with members from machine learning, neuro imaging, neuropathology, and computational neuroscience. Your interactions will include close ties to clinicians for feedback, collaboration, and the development of ideas. Students from these fields will be in the same lab.   
  • Cutting-edge computing environment including large-scale HPC/GPU cluster for deep learning

Who you are: 

  • Master in Computational Neuroscience, Machine Learning or related fields
  • Experience and interest in any of machine learning, medical image analysis, computational neuroscience, or network analysis 
  • Motivated to develop novel machine learning approaches in the field of medical imaging and precision medicine. 

Foreseen secondments:: 

  • Dr. Meritxell Bach Cuadra, University of Lausanne (3 months); advanced network connectivity analysis, change tracking and modelling of dynamic network characteristics; 
  • Prof. Roland Wiest, University of Bern (3 months), Clinical brain imaging and ML approaches; 
  • Dr. Dirk Smeets, CTO at Icometrix,(1 month), research code into production code in a clinical software, state-of-the-art brain processing methodology.
Apply now!

Employer: University of Lausanne, Lausanne, Switzerland 

PhD enrollment: Life Sciences Program, University of Lausanne, Lausanne, Switzerland 

Main supervisor: Prof. Meritxell Bach Cuadra 

PhD project description: Multiple sclerosis (MS) is a complex neurodegenerative disease traditionally diagnosed through intermittent clinical episodes and longitudinal MRI monitoring of brain lesions and atrophy. However, this binary classification approach—relapsing vs. remitting, active vs. inactive—fails to capture the continuous spectrum of brain degradation increasingly revealed by recent biopathological research. Current monitoring methods rely on limited observational markers that cannot explain disease mechanisms or predict individual trajectories. This PhD project aims to transform MS characterisation from categorical labels to continuous quantification by developing a Deep Structural Causal Model (DSCM) framework that integrates observational data with expert clinical knowledge to uncover disease mechanisms. 

First, the project will tackle the prevalent problem of technical variability in clinical brain MRI. Images acquired across different scanner vendors, protocols, contrasts, and resolutions create distribution shifts that introduce errors in automated biomarker extraction, requiring costly and unreliable model retraining. The candidate will explore two complementary state-of-the-art approaches: (a) SO(3)-equivariant deep learning models capable of handling varying resolutions across different scanner field strengths (0.5T–7T) (in co-supervision with Dr Richard McKinley Inselpital Bern), and (b) brain MRI foundation models with rapid task adaptation capabilities. A key innovation will be investigating how incorporating the longitudinal dimension can enhance both approaches and create synergies between them. 

Second, the project will develop a comprehensive causal framework for MS. Derived imaging biomarkers, alongside multimodal data (omics, demographics, clinical assessments), will be integrated into a SCM guided by clinical expert knowledge. This SCM will provide an explicative skeleton of the MS data generation process (from underlying pathology to observable imaging features and diagnosis), enabling mathematical characterisation of disease mechanisms through causal theory. While traditional causal models require abundant observations, longitudinal imaging studies face inherent data limitations. The candidate will also explore other advanced methods (e.g., GLACIAL) to address this constraint. 

Expected outcomes include: 

  • Novel equivariant and foundation model architectures robust to multi-site, multi-contrast, and multi-resolution MRI data 
  • A DSCM framework linking imaging biomarkers to disease mechanisms through causal relationships, validated against clinical expert knowledge 
  • Methodological advances for causal inference from limited longitudinal imaging data 
  • Open-source modules integrating longitudinal constraints for improved MS characterisation 
  • Validated tools for continuous MS quantification that move beyond binary disease categorisation 

What we offer

  • A multidisciplinary project between cutting-edge brain imaging and advanced image processing, machine learning, and clinical applications.  
  • A dynamic, interdisciplinary, and international team of very motivated people, including collaborations with MS clinical experts. 
  • A stimulating work environment and access to cutting-edge technology and state-of-the-art resources.  

Who You Are   

  • A master’s (MSc) degree in physics, computer science, or electrical engineering, or similar degree with an equivalent academic level.  
  • You have experience in machine learning and deep learning with applications in medical imaging.   
  • A strong will to develop clinically actionable methods and to interact with clinicians is required.  
  • Good programming skills Python, including full stack and deep learning frameworks (PyTorch or TensorFlow).  
  • Experience with brain MRI data, or longitudinal analysis or causal theory is a plus but not mandatory.  
  • Good English skills (oral and written) are required.  

Foreseen secondments: For this PhD project, research visits are foreseen to:   

  • Prof. Cristina Granziera (3 months) at University Hospital Basel, Switzerland  
  • Dr. Bénedicte Maréchal (3 months) at Siemens Healthineers, Lausanne, Switzerland  
  • Prof. Pietro Maggi (4 months) at Université Catholique de Louvain, Belgium
Apply now!

Employer: Aalto University, Espoo, Finland

PhD enrollment: Aalto University, Espoo, Finland

Main supervisors: prof. Koen Van Leemput & Dr. Richard McKinley (Bern, Switzerland)

Clinical Problem: Among patients who present with a first seizure, approximately 45% will have a further seizure within the next five years. Identifying which patients are likely to have recurrent seizures is therefore vitally important, since early treatment can dramatically improve the lives of patients with epilepsy and spare them from further neurological deficits brought on by repeated seizures.

Structural brain lesions are the most common cause of epileptic seizures, but the mere presence of a lesion is not predictive of epilepsy: the size and location of the lesion, and its proximity to the grey matter, are critical factors. Similarly, while loss of grey matter is seen in all epilepsy patients over time, specific patterns of grey-matter atrophy can be associated to different epilepsy subtypes. Identifying these patterns in a timely manner has the potential to drive patient treatment.

Your solution: In this PhD project you will develop methods to link patterns of grey- and white-matter alterations to epilepsy subtypes and to distinguish epilepsy patients from those with epilepsy mimics. You will benchmark and extend existing segmentation methods for quantifying white matter lesions and grey matter atrophy, and establish baselines for linking alterations and disease, such as lesion-symptom mapping and structural covariance networks. You will then explore how generative modeling can be harnessed to translate patterns of brain alteration into low-dimensional latent codes which capture the essence of a patient’s deviation from normal brain appearance, and harness these codes to better estimate a patient’s risk of receiving a definitive epilepsy diagnosis. The methods developed will be based on data from, and have direct application in, ongoing studies at the University Hospital Bern and across Switzerland, which aim to identify patients at high risk of developing epilepsy after a first seizure, and to study the effects of epilepsy over time on the structure, function and metabolism of the brain.

Foreseen secondments: For this PhD project, research visits are foreseen to:

  • Prof Roland Wiest & Dr. Richard McKinley, Inselspital, University Hospital Bern, Switzerland (10 months across 3 visits)
  • Dr Ankur Sharma, Bayer (Berlin, Germany) (2 months)
Apply now!

Employer: University of Bern, Switzerland

PhD enrollment:  University of Bern, Switzerland

Main supervisor: prof. Roland Wiest 

Clinical problem: Patients with an epileptic seizure frequently present with unspecific neurologic symptoms mimicking other conditions, such as stroke, migraine or brain tumour. Imaging biomarkers are therefore crucial to distinguish subtle seizures from other conditions in the emergency setting. Subtle abnormalities in cerebral blood flow, which occur during and after seizures, are one promising biomarker: however, current methods for the processing of perfusion MRI data are sensitive to image noise, preventing reliable depiction of these changes in individual patients.

Your solution: In this PhD project you will develop data-driven methods to improve the processing of perfusion MRI data, building novel spatio-temporal denoising methods based on diffusion models that allow subtle perfusion-related signal to be detected. These enhanced perfusion signals will then be leveraged to distinguish between epilepsy and epilepsy-mimics and to identify epilepsy subtypes. The methods developed will have direct application in ongoing studies at the University Hospital Bern and across Switzerland, which aim to identify patients at high risk of developing epilepsy after a first seizure, and to study the effects of epilepsy over time on the structure, function and metabolism of the brain.

Our Team
You will be hosted in the Support Centre for Advanced Neuroimaging (SCAN), a multidisciplinary imaging laboratory based at the Inselspital, University Hospital Bern. The SCAN unites medical doctors, physicists, biomedical engineers and computer scientists into a multidisciplinary team to develop and clinically validate advanced neuroimaging technologies.

Foreseen secondments: For this PhD project, research visits are foreseen to: 

  • Prof. Koen Van Leemput at Aalto University, Espoo Finland (3 months)
  • Dr. Tom Hilbert at Siemens Healthineers, Lausanne (2 months)
  • Prof. Georg Langs, Medical University of Vienna, Austria (1 month)
Apply now!

FAQ page with more information

Recruitment privacy notice