By leveraging theories and methodologies in computer vision, machine/deep learning, medical imaging, and human-computer interaction, the research at Health-X aims to help improve the accuracy and efficiency of clinical diagnoses and medical procedures. The applications include neurodegenerative conditions (e.g., Parkinson's disease), neurovascular diseases, brain cancer, and musculoskeletal disorders. 

---

Computational modelling of anatomy and diseases

Various diseases and natural aging can result in characteristic changes to the anatomy, cellular tissue properties, and functional measures (e.g., brain connectivity). With medical imaging data, such as MRI and EEG, we want to capture these insights with data-efficient (weak and self-supervision, few-shot learning) and explainable AI techniques, and employ tailored computational models to understand and predict their impacts and progression for patients. This will be instrumental for the management of diseases, testing new treatments, and fundamental understanding of our physiology.

Immersive & Augmented health technology

Virtual and Augmented Reality can allow more natural visualization and interaction of complex digital data, as well as simulate various 3D environments and scenarios. We want to explore novel, intuitive, and efficient human-computer-interaction methodologies and multi-modal data for virtual/augmented reality to facilitate the planning and monitoring of medical procedures, deliver effective medical educational materials, and provide accessible tools for improving physical and mental health.

Medical human-AI collaboration

Various human factors can affect the quality of clinical diagnosis and procedures, especially when it comes to perceiving and interpreting medical scans. Although machine/deep learning techniques have shown great potential in radiological applications, the future of computer-assisted diagnosis and procedures can greatly benefit from a user-centered design involving collaboration of clinicians and the machine. This paradigm will help enhance the reliability, safety, and efficiency of computer-assisted clinical decisions while improving the skill training of both clinicians and AI algorithms. Specifically, we want to better understand the influence of human factors, integrate human behavioural data (e.g., gaze), and devise new methods (e.g., multi-modal learning, foundation models, AI agents) to enable intuitive human-computer interaction in clinical tasks and training.