Hongwei Li (Bran) 李宏伟
- Email: hongwei.li AT tum.de.
- Office: First floor of IMETUM, Boltzmannstr. 11, 85748, Garching.
My research interests focus on developing basic generalizable and efficient machine learning algorithms in the field of medical image analysis.
Previously, I was supervised by Prof. Jianguo Zhang and Prof. Wei-Shi Zheng during my master's program in China (Guangzhou) and the UK (Dundee).
Research
My program includes several cool projects collaborating with clinical partners in Munich and overseas:- Domain adaptation/generalisation for brain lesion segmentation;
- Medical image synthesis and reconstruction of brain and spine using generative adverarial network;
- Robust cardic structure segmentation;
- Whole brain segmentation in MR images;
- Motion artefact removal in k-space data.
Selected Publications
12. Imbalance-aware self-supervised learning for 3d radiomic representations, MICCAI 2021 (06/2021)11. Automated claustrum segmentation in human brain MRI using deep learning, Human Brain Mapping
10. Domain Adaptive Medical Image Segmentation via Adversarial Learning of Disease-Specific Spatial Patterns, under review
9. Deep-Learning Generated Synthetic Double Inversion Recovery Images Improve Multiple Sclerosis Lesion Detection, Investigative Radiology (10/2019)
8. Adversarial Convolutional Networks with Weak Domain-Transfer for Multi-Sequence Cardiac MR Images Segmentation, MICCAI-STACOM (08/2019)
7. DiamondGAN: Unified Multi-Modal Generative Adversarial Networks for MRI Sequences Synthesis, MICCAI 2019 (06/2019)
6. Differential Diagnosis for Pancreatic Cysts in CT Scans Using Densely-Connected Convolutional Networks, EMBC (02/2019)
5. Fully Convolutional Network Ensembles for White Matter Hyperintensities Segmentation in MR Images, NeuroImage (08/2018)
4. Multi-Scale Convolutional-Stack Aggregation for Robust White Matter Hyperintensities Segmentation, MICCAI-BrainLes Workshop (08/2018)
3. Automatic Brain Structures Segmentation Using Deep Residual Dilated U-Net, MICCAI-BrainLes Workshop (08/2018)
2. HEp-2 specimen classification via deep CNNs and pattern histogram, International Conference on Pattern Recognition (ICPR 2016)
1. HEp-2 cells staining patterns classification via wavelet scattering network and random forest, Asian Conference on Pattern Recognition (ACPR 2015)
Projects for Bachelor or Master Students
I enjoy working with self-motivated students who are interested in machine learning and medical image analysis.Please feel free to contact me and Prof. Bjoern Menze if you are keen to do research on relevant topics as a part of your study program.
Open-Source Codes (continuously updated)
- 1) WMH segmentation algorithm, winning method for MICCAI WMH Segmentation Challenge 2017
- 2) DiamondGAN for multi-modal MRI sequences synthesis
- 3) brain structures segmentation, a top-performing 2D-based approach in MR brain segmentation challenge.
Other Interests
Amateur indoor and beach volleyball player, once played for VfR Garching Men's volleyball team in Landesliga for one season;Reading and writing fairy tales.
updated on 01/08/2021
