معرفی
Irina Voiculescu is an Associate Professor and Stipendiary Lecturer in the Department of Computer Science at the University of Oxford. Her research focuses on 3D geometric modeling, medical imaging analysis, and AI-driven clinical decision support. She leads the Medical Imaging research group, developing tools like OxMedIS for automated segmentation of MRI, CT, and ultrasound data. Her work emphasizes GPU-accelerated algorithms and reinforcement learning to reduce reliance on manual annotations.
Education: PhD in Constructive Solid Geometry (CSG) from the University of Bath (1999). Research interests include medical image segmentation, GPU computing, cephalometric landmark detection, and infant hip dysplasia screening. Collaborates with orthopaedic clinicians and industry partners.
Notable contributions: Developed novel evaluation metrics for segmentation accuracy, pioneered GPU-based watershed algorithms, and advanced AI applications in hip and knee imaging. Her team’s work on automated Graf classification for infant hip screening was recognized at ECCV 2024.
Awards: Best Paper Award at DEMI 2023, Best Workshop Paper at HSI 2009. Editorial roles include Associate Editor for SPIE Journal of Medical Imaging.
Advising: Supervises PhD students in medical AI, including NIHR-funded researchers. Past students include Nanqing Dong (federated learning) and Ziyang Wang (transformer networks). Runs the Spatial Reasoning group, known for the Oxford Robot Games and robotic sheepdog projects.
Publications: Over 80 peer-reviewed papers since 2008, with recent focus on few-shot learning, domain adaptation, and clinical validation of AI models.
