Roger Pallares Lopez

Roger Pallares Lopez is a graduate student in mechanical engineering merging computer vision with physics-based simulation to extract tissue motion from ultrasound without manual annotation. Ultrasound captures rich information about musculoskeletal dynamics during movement, yet analyzing data at scale remains impractical due to inherent ultrasound image noise, anatomical complexity, and variability introduced by acquisition conditions. Roger adapted vision foundation models to ultrasound and developed MATLAB pipelines using k-Wave and the Signal Processing Toolbox to generate synthetic B-mode sequences with known motion and realistic speckle patterns. Models trained on synthetic data have shown encouraging early tracking results on real ultrasound sequences, supporting the potential of simulation-based supervision for ultrasound motion analysis. As a second-year MathWorks Fellow, Roger will accelerate data generation through scientific machine learning and generative modeling approaches. His contributions could enable widespread adoption of ultrasound-based tissue analysis in clinical and research settings where traditional annotation is prohibitively expensive.