A5010908329- Cell Image Analysis Techniques
- Cytomegalovirus and herpesvirus research
- Electron and X-Ray Spectroscopy Techniques
- Advanced Electron Microscopy Techniques and Applications
- Image Processing Techniques and Applications
- Herpesvirus Infections and Treatments
- Domain Adaptation and Few-Shot Learning
- Machine Learning in Materials Science
- AI in cancer detection
- Radiomics and Machine Learning in Medical Imaging
- Cancer Genomics and Diagnostics
- Ideological and Political Education
- Ocular Diseases and Behçet’s Syndrome
Universität Ulm
2018-2022
Linköping University
2014
Semantic segmentation of electron microscopy images using deep learning methods is a valuable tool for the detailed analysis organelles and cell structures. However, these require large amount labeled ground truth data that often unavailable. To address this limitation, we present weighted average ensemble model can automatically segment biological structures in when trained with only small dataset. Thus, exploit fact combination diverse base-learners able to outperform one single model. Our...
Detailed analysis of secondary envelopment the herpesvirus human cytomegalovirus (HCMV) by transmission electron microscopy (TEM) is crucial for understanding formation infectious virions. Here, we present a convolutional neural network (CNN) that automatically recognises cytoplasmic capsids and distinguishes between three HCMV capsid stages in TEM images. 315 images containing 2,610 expert-labelled classes were available CNN training. To overcome limitation small training datasets thus poor...
Image processing algorithms in pathology commonly include automated decision points such as classifications. While this enables efficient automation, there is also a risk that errors are induced. A different paradigm to use image for enhancements without introducing explicit Such can help pathologists increase efficiency sacrificing accuracy. In our work, has been applied Ki-67 hot spot detection. scoring routine analysis quantify the proliferation rate of tumor cells. Cell counting spot,...
Scanning Transmission Electron Microscopes (STEMs) acquire 2D images of a 3D sample on the scale individual cell components. Unfortunately, these can be too noisy to fused into useful structure and facilitating good denoisers is challenging due lack clean-noisy pairs. Additionally, representing detailed difficult even for clean data when using regular grids. Addressing two limitations, we suggest differentiable image formation model STEM, allowing learn joint sensor noise in STEM together...
Scanning Transmission Electron Microscopes (STEMs) acquire 2D images of a 3D sample on the scale individual cell components. Unfortunately, these can be too noisy to fused into useful structure and facilitating good denoisers is challenging due lack clean-noisy pairs. Additionally, representing detailed difficult even for clean data when using regular grids. Addressing two limitations, we suggest differentiable image formation model STEM, allowing learn joint sensor noise in STEM together...