AI-informed adaptive mid-field MRI for personalized treatment of depression

Depression affects many people, but antidepressant treatment does not work equally well for everyone. Predicting treatment outcomes early could help clinicians make more informed decisions and tailor treatment to individual patients.

In this postdoc project, you will investigate how adaptive, AI-informed mid-field MRI, guided by questionnaire data, can provide clinically useful biomarkers for predicting antidepressant treatment outcomes. By combining artificial intelligence, brain imaging and digital mood and symptom data, you will work towards a more affordable and widely accessible approach to precision psychiatry.

You will focus on three main areas:

• Exploring how adaptive mid-field MRI can reduce scan time and improve prediction performance compared with fixed scanning protocols, while remaining feasible in outpatient clinics and settings with limited resources.
• Designing an MRI-based prediction pipeline using perfusion measurements from relevant brain regions, acquired with arterial spin labelling (ASL).
• Developing adaptive MRI acquisition methods in which AI models, including reinforcement learning, determine what to scan next based on the information already collected.

You will build on data from a large randomized clinical trial on treatment outcomes in major depressive disorder. Using these data, you will adapt existing prediction models and investigate whether mid-field perfusion MRI can support the early prediction of antidepressant treatment outcomes.

You will develop an end-to-end prediction pipeline and translate your methods into an adaptive MRI strategy that can run directly on the scanner. You will also design and conduct a pilot MRI study in healthy volunteers, testing a prototype model deployed on the scanner.

In addition, you will share your findings with the scientific community and contribute to the further development of the PRESS platform in collaboration with consortium partners from academia and industry. Your responsibilities include:

• Adapting existing predictive models to explore the feasibility of mid-field perfusion MRI for predicting antidepressant treatment outcomes.
• Developing an AI-based prediction pipeline that combines MRI perfusion measurements with relevant clinical and digital data.
• Implementing and testing adaptive mid-field MRI acquisition strategies, including a prototype model deployed on the scanner.
• Designing and conducting a pilot MRI study in healthy volunteers.
• Writing scientific papers, presenting your findings at international conferences and contributing to the integration of your methods into the PRESS platform.

You have a PhD in artificial intelligence, computer science, (biomedical) engineering, physics, computational neuroscience, technical medicine or a closely related field. You enjoy developing new technical approaches and applying them to clinically relevant challenges.

In addition, you bring:

• Strong programming skills, for example in Python, PyTorch or TensorFlow, and experience with modern machine learning workflows.
• An affinity with, and preferably experience in, one or more of the following areas: MRI or other medical imaging, reinforcement learning or sequential decision-making algorithms, and clinical prediction modelling or health data science. Experience with perfusion imaging or ASL is an advantage.
• A clear interest in mental health, neuroimaging and translational research at the intersection of AI and medicine.
• Excellent written and spoken English, good communication skills and the ability to work independently and proactively in a multidisciplinary research environment.