Welcome to my GitHub page! My experience spans computer vision, natural language processing, graph machine learning, and time-series forecasting. I am particularly interested in online learning, sequence modeling with transformers, video prediction, and generative AI. I currently work at Brainomix (Oxford, UK), where I explore machine learning for computer-aided diagnosis of interstitial lung diseases.
| Chest and liver cine-MRI prediction using PCA-based motion modeling and temporal dynamics forecasting with RNNs trained online and transformers | Deformable 3D image registration using the pyramidal, iterative Lucas–Kanade optical-flow algorithm |
|---|---|
|
6-step-ahead prediction of sagittal cine-MRI cross-sections using sparse one-step approximation (left: ground truth, right: prediction).  |
Estimation of breathing-related 3D lung-tumor motion using deformable image registration.  |
| Time-series forecasting using RNNs trained online and transformers | |
|
Prediction of the 3D positions of 3 markers on the chest of a subject 2.1s in advance (i.e., 7 time steps ahead, at a sampling rate of 3.33Hz) using RNNs trained with decoupled neural interfaces, enabling latency compensation during lung radiotherapy treatment.  |
|