Quantifying hygroscopic deformation in lignocellulosic tissues: a digital volume correlation tool comparison
Evaluate DVC methods on nanoCT data of lignocellulosic tissues.
kim-ulrich
Professional Summary
Tak Ming Wong is a research scientist at Helmholtz-Zentrum Hereon, Germany. His research interests are mainly in computational imaging, computer vision, scientific machine learning (SciML) for advanced imaging modalities.
Interests
Computational Imaging
Computer Vision
Scientific Machine Learning (SciML)
Knowledge-guided Deep Learning
Advanced Imaging Modalities (CT, THz)
AI
📚 About me
I’m a research scientist at the Hereon’s outstation located at the largest particle accelerator in Germany (DESY, Hamburg). Currently, I’m a member of Imaging and Data Science Department of the Institute of Metallic Biomaterials and a member of the X-Ray Imaging with Synchrotron Radiation Department of the Institute of Materials Physics.
Prior to this, I completed my doctoral research at the Computer Graphics and Multimedia Systems Group under the supervision of Prof. Andreas Kolb and the co-supervision of Prof. Peter Haring Bolívar at the University of Siegen in 2023.
I develop deep neural networks to solve inverse problems in tomography imaging (at Hereon’s μCT and nanoCT beamlines) for materials science experiments. Currently, I’m particularly focusing on memory efficiency of volumetric networks and knowledge-guided deep learning methods for downstream tasks such as:
- Digital Volume Correlation (DVC)
- Phase retrieval
- Flat-field correction
Please feel free to reach out for collaboration and opportunities! 😃
Selected Papers
•
•
1 min read
Self-supervised physics-informed generative networks for phase retrieval from a single X-ray hologram
SelfPhish: self-supservised and physics-informed GANs for phase retrieval.
xiaogang-yang
VolRAFT: Volumetric Optical Flow Network for Digital Volume Correlation of Synchrotron Radiation-based Micro-CT Images of Bone-Implant Interfaces
VolRAFT: extend optical flow network RAFT to volumetric data for digital volume correlation (DVC).
Deep Optimization Prior for THz Model Parameter Estimation
Deep Optimization Prior approach allows to find better local optima in the non-convex energy landscape.
Training Auto-Encoder-Based Optimizers for Terahertz Image Reconstruction
An unsupservised model-based autoencoder in which the encoder network predicts suitable parameters and the decoder is fixed to a physical model.
Papers
(2026).
An ontology-based description of nano computed tomography measurements in electronic laboratory notebooks.
Scientific Data.
(2025).
Quantifying hygroscopic deformation in lignocellulosic tissues: a digital volume correlation tool comparison.
Frontiers in Plant Science.
(2025).
Self-supervised physics-informed generative networks for phase retrieval from a single X-ray hologram.
Optics Express.
(2025).
A Framework for the AI-based visualization and analysis of massive amounts of 4D tomography data for end users of beamlines.
14th Conference on Industrial Computed Tomography (iCT) 2025.
(2024).
Machine Learning for the Reconstruction and Analysis of Synchrotron-radiation Tomography Data.
Optical Engineering + Applications, 2024, San Diego, California, United States.
Recent News
Recent & Upcoming Talks