Match4Annotate
Transferring point and mask annotations across independently acquired ultrasound videos without target-side labels.
View work
Roger Pallares Lopez
PhD Candidate at MIT | Computer Vision · Medical Imaging · Machine Learning for Health
I build computer vision and machine learning methods for complex health data, with applications in medical imaging, cardiology, and human movement. My work focuses on turning high-dimensional, multimodal measurements into reliable and clinically relevant insights.
About
I am a PhD candidate in Mechanical Engineering at MIT, where I develop computer vision and deep learning tools for health and medicine. My current research focuses on efficiently propagating annotations across images and video frames, as well as tracking anatomical motion in ultrasound videos to extract dynamic measures that can inform clinical assessment and our understanding of health and well-being. I also develop machine learning methods for problems in cardiology and human movement, often using complex, multimodal data. Complementing this work, I hold a master’s degree in control and robotics, which brings a perspective on dynamics, control, and embodied systems to my training in vision and learning. Throughout my work, I aim to contribute to solutions that are reliable in real-world data and produce insights that researchers and clinicians can interpret and use.
News
- I received the MathWorks Fellowship (opens in a new tab) for a second year, funding another year of my PhD studies at MIT.
- Our paper, DUSTrack: Semi-automated point tracking in ultrasound videos (opens in a new tab), was published in Scientific Reports.
- Our multimodal biomechanics dataset (opens in a new tab) with synchronized kinematics and internal tissue motion was published in Scientific Data.
