A Ph.D. position is available in our group to contribute to the development of efficient, certifiable, and practical solutions for inverse problems in diagnostic biomechanics, with a particular focus on elastography. The project builds on our recently developed Weak Neural Variational Inference
(WNVI) framework and aims to advance its capabilities in the following directions:
• Scalability: Extending methods to high-dimensional and three-dimensional elastography
problems using neural operator representations.
• Computational efficiency: Designing adaptive and physics-aware strategies (e.g., optimized residual selection, physics-based zooming) for real-time inference.
• Practical usability: Developing robust, user-friendly frameworks for multimodal elastography and enabling deployment on portable devices (e.g., smartphones) for real-time diagnostics.
The research combines continuum mechanics, machine learning, computational mathematics, and probabilistic modeling, with direct applications in medical imaging and beyond.
More information can be found here.