Yordanka (Dani) Velikova

I am a PhD researcher in the Computer-Aided Medical Procedures Lab (CAMP) at the Technical University of Munich, under the supervision of Prof. Navab where I work on ultrasound image segmentation and robotic ultrasound.

Email  /  Google Scholar  /  Twitter  /  Github

Research

My research is primarily focused on unsupervised medical image segmentation in particular for ultrasound imaging as well as image-based robotic ultrasound navigation. Beyond this, my interests branch into implicit neural representations, spectral embedding, shape completion, breathing compensation, and object tracking with transformers.

A physics-inspired diffusion model for ultrasound image generation, simulating sound wave propagation for realistic ultrasound image synthesis.

A novel deep spectral method for unsupervised interpretation of ultrasound images, advancing segmentation techniques through spectral clustering.

This project addresses the challenge of completing vertebrae morphology from 3D ultrasound imaging, using novel shape completion techniques even in low-visibility or incomplete datasets.

This project focuses on using implicit neural representations to reconstruct 3D volumes from robotic ultrasound while compensating for respiratory motion during scanning.

Optimizing Ulrasound Intermediate Representations guided by the segmentation task. Trained end-to-end with domain adaptation network to account for the domain gap between real and simulated.

Common anatomical space between CT and US is an Intermediate Representation (IR) which acts as a virtual third modality. It inherits properties from both CT and Ultrasound and preserves the patient-specific anatomical layout.

MICCAI Journal version: version / bibtex

Template webpage from source code.