Publications
Diagnostic Accuracy of an Automated Classifier for the Detection of Pleural Effusions in Patients Undergoing Lung Ultrasound
AI that detects fluid around the lungs on ultrasound, tuned for different clinical settings.
LangDA: Language-guided Domain Adaptive Semantic Segmentation
Adapting image segmentation models to new domains using language descriptions of scenes.
Implicit Stylization for Domain Adaptation
Adapting models to new domains via lightweight feature alignment instead of costly image stylization.
Building Spatio-temporal Transformers for Egocentric 3D Pose Estimation
Transformers that use past video frames to estimate 3D body pose from head-mounted cameras.
In the Press
Open models have real benefits. Do they outweigh the risks?
“My reason for optimism is that it’s a very active area of research, and a lot of people are trying to come up with new and better methods.”
Major security weaknesses found in leading open AI models
“The defences available today don’t yet appear strong enough to guarantee that a publicly released model will remain safe once it’s in the hands of anyone who chooses to modify it.”
Tests raise fears open AI models could be ‘used at scale for harm’
BME undergraduate student received Honorable Mention for Outstanding Undergraduate Researcher Award
Talks & Panels
Translational Pathways in Biomedical Engineering: From Lab to Clinic
Augmented Reality for Guided Lung Ultrasound Acquisition
Enhancing AI Testing: The Power of Metadata Hydration
Mastering RAG Systems for LLMs: From Pitfalls to Performance
