Irina Voiculescu

Irina Voiculescu is an Oxford computer scientist whose AI work develops methods to analyze medical images for clinical use. She joined Oxford’s Department of Computer Science in 1999; her PhD is in 3D geometric modelling from the University of Bath, and she later coauthored a report on evaluating medical image segmentation. Her research includes landmark measurements from medical scans and a scribble-based method for cardiac MRI segmentation; the latter was runner-up for the 2025 MIUA Best Paper Award. In 2026, she coauthored a perspective on uncertainty in medical segmentation datasets. As of October 2026, she is an Associate Professor at Oxford and leads the Oxford Medical Imaging Science group.

Now at: Oriel College, University of Oxford

Timeline

Since 1999

Joining Oxford computer science

She joins the University of Oxford’s Department of Computer Science in 1999. Her PhD is in 3D geometric modelling from the University of Bath.
2001

Oxford’s robot flock

The Spatial Reasoning group builds a robot sheepdog to herd live or robotic animals; Oxford’s robotic flock appears at the Royal Society Summer Science Exhibition in 2001.
2015

A segmentation metrics report

In 2015, she coauthors the Oxford technical report “An Overview of Current Evaluation Methods Used in Medical Image Segmentation” with Varduhi Yeghiazaryan. The report reviews methods for assessing medical image segmentation.
May 2024

Landmarks in medical scans

At a STAR-UBB event in May 2024, she gives an invited lecture titled “Distance and Angle Measurements from Landmarks in Medical Images.” Its abstract describes contour detection and landmarks as tools for deriving clinical measurements from scans.
2025

Cardiac MRI from scribbles

In 2025, she coauthors “Semantic Segmentation with Spreading Scribbles” with Varduhi Yeghiazaryan and Yeva Gabrielyan. The paper is runner-up for the MIUA Best Paper Award and shows that rough clinician scribbles can guide segmentation of the myocardium and ventricles in cardiac MRI.
2026

Uncertainty in medical segmentation

In 2026, she coauthors a perspective with Sophie Fischer, “The hidden value of uncertainty in medical segmentation.” It examines how medical image datasets represent uncertainty in annotations made by clinicians.
As of October 2026

Associate Professor at Oxford

As of October 2026, she is Associate Professor of Computer Science at Oxford and Stipendiary Lecturer at Oriel College. She leads the Oxford Medical Imaging Science (OxMedIS) group, which works on segmentation, 3D reconstruction, anatomical landmark detection, multimodal diagnosis and reliable healthcare AI.

Career

NowUniversity of OxfordAssociate Professor of Computer Science, Oxford Medical Imaging Science (OxMedIS) Group Leader
NowOriel College, University of OxfordStipendiary Lecturer in Computer Science