Image Analysis
Our image analysis team focuses on the standardized and exploratory analysis of tissue images obtained with fluorescent immunohistochemistry. Additionally, we process conventional H&E and chromogenic IHC images.
Image Analysis
Our image analysis team focuses on the standardized and exploratory analysis of tissue images obtained with fluorescent immunohistochemistry. Additionally, we process conventional H&E and chromogenic IHC images.
We divide our analysis into three pillars:
State-of-the-art “out-of-the-box” analysis with commercial software.
This includes tissue segmentation, cell segmentation, cell phenotyping and subsequent quantification and statistical analysis to answer predefined hypothesis and to get a general baseline overview of the data.
Spatial analysis and custom feature extraction.
This includes hypothesis-driven analysis and exploratory questions whenever the image feature of interest goes beyond simple cell quantification. This may include detection and measurement of custom image objects, distance measurements between predefined cell-phenotypes, identification of touching cells and mutual nearest neighbors, and the incorporation of manual annotations of objects and tissue areas.
Machine learning and synthetic image data.
For larger and more complex research questions, we develop custom analysis pipelines based on deep learning algorithms for predictive models, object identification and image classification. This includes the generation of synthetic image data based on real samples using both procedural and machine-learning-based methods. Knowledge extraction pipelines are used to explain the reasoning of trained machine learning models to understand model decision making and to infer new insights about the data.