PhD
Computer Aided Medical Procedures and Augmented Reality at TUM
Boltzmannstraße 15, 85748
Garching bei München
Biosketch
Miruna is a Ph.D. student at the Technical University of Munich (TUM) in the Chair of Computer Aided Medical Procedures and Augmented Reality. She holds a Master’s degree in Biomedical Computing from TUM, where she researched 3D Shape Completion for Ultrasound-based Spine Injections for her thesis. Her research focuses on improving ultrasound interpretability and visualization, aiming to advance medical technology and enhance patient outcomes.
relAI Research
Deep Learning for improving interpretability and visualization in ultrasound imaging
My research develops deep learning methods that make clinical ultrasound easier to interpret. Unlike MRI or CT, which provide high-contrast volumes, ultrasound is constrained by restricted acoustic windows and artefacts such as reverberations, acoustic shadowing, and impedance mismatches that lead to occlusions and ambiguities. I design neural networks that compensate for these limitations by integrating anatomical prior knowledge and physics-inspired inductive biases within cross-modal learning frameworks. I focus on completion models that infer hidden anatomy beyond the transducer’s field of view, reconstructing plausible 3D context from sparse 2D slices. I further extend these methods to robotic ultrasound systems that guide probe placement toward optimal acoustic windows and flag unreliable image regions. Together, these contributions aim to transform ultrasound from a user-dependent, artefact-prone modality into a transparent, interpretable imaging tool.
Publications
https://doi.org/10.1007/s11548-024-03126-x
https://doi.org/10.1109/IROS60139.2025.11247644
https://doi.org/10.1007/978-3-032-06774-6_17