Treatment planning for transcranial ultrasound therapy using deep learning

Lead Research Organisation: University College London
Department Name: Institute of Health Informatics


Project lay summary
The project looks at obtaining accurate and personalised maps of the skull geometry and material properties from MR/CT images. This can accelerate the prediction of the transcranial acoustic field and allow for real-time treatment planning. In addition, it includes generating a statistical shape model of the skull, developing tailored deep learning architectures, and analysing model uncertainty and interpretability. Finally, the models will be rigorously tested using patient data from previous clinical treatments.

Supervisor details:
Professor Bradley Treeby (
Professor Ben Cox (
Dr Antonio Stanziola (,
Department of Medical Physics and Biomedical Engineering, UCL

Brief description of the context of the research including potential impact
Brain disorders affect as many as one third of the adult population. One exciting treatment alternative to drugs and surgery is the use of ultrasound. Different applications include precise destruction of small regions of tissue, transcranial ultrasound stimulation (TUS) and temporary opening of the blood-brain barrier. Clinical trials in the last few years have demonstrated that these techniques can be highly effective. A major challenge for ultrasound therapy is ensuring the ultrasound energy is delivered to the precise location identified by the clinical team, raising the need to predict the distortions of the waves caused by the skull bone. However, existing computer models take many hours or days to run and require subjects to have a CT scan which is undesirable for healthy subjects. These challenges currently limit the application of ultrasound therapy, particularly in the deep brain.

Aims and objectives
The main goal is to leverage deep learning to deliver personalised treatments for patients undergoing ultrasound therapy in the brain.
Particularly the objective is to obtain an accurate representation of the acoustic properties of the skull in order to plan a precise treatment.

The research methodology, including new knowledge or techniques in engineering and physical sciences that will be investigated
Neural network architectures tailored to the above tasks

Alignment to EPSRC's strategies and research Areas:
Artificial intelligence technologies (

Any companies or collaborators involved:
Centro Integral en Neurociencias A.C. (HM CINAC) in Spain
Physiological Neuroimaging Group at the University of Oxford


10 25 50

Studentship Projects

Project Reference Relationship Related To Start End Student Name
EP/S021612/1 31/03/2019 29/09/2027
2418780 Studentship EP/S021612/1 27/09/2020 29/09/2024 Maria Miscouridou