Strand Lead: Professor Natalio Krasnogor
Immunohistochemistry (IHC) plays an essential role in molecular pathology, and quantification of IHC can be achieved through whole slide imaging (WSI). The idea of WSI is to use digital microscope scanners to capture the glass slides as digital slide images. The analysis of immunohistochemistry digital images can help pathologists with disease diagnosis, provides key information for understanding the disease, and is essential for the validation of in vivo biomarkers that allow the stratification and monitoring of patients at risk.
One of the major challenges to analyse immunohistochemistry images is the development of computational image analysis methods. Manual or semi-automated methods are time-consuming and require user intervention with an element of subjectivity and inter-observer variability. Image analysis approaches commonly used for quantifying immunohistochemistry data that relies on out-of-box software, such as ImageJ or CellProfiler, struggle with inhomogeneities and complexities in immunohistochemistry images. Most digital microscope companies sell image analysis software. However, the cost of those software are high, and they only provide limited image processing capabilities.
The mission of the Node’s computing strand is to develop tools to provide high throughput automatic image analysis of immunohistochemistry images, to gain insight into the biology and function of cells in health and disease, using computer vision methods such as image enhancement, image segmentation, image feature extraction and classification.
Furthermore, we will supplement automated image analysis with advanced machine learning methods to analyse expression data, perform functional analysis of biological (e.g. transcription, signalling, metabolic, etc) networks, inference of missing links in regulatory pathways, genes-set and network based enrichment analysis, etc.