frcoloc

Colocalization of Fluorescence and Raman Microscopic Images for Training Data Collection

This colocalization scheme unveils statistically significant overlapping regions by identifying correlation between fluorescence color channels and clusters from unsupervised machine learning methods like hierarchical cluster analysis (HCA) performed on Raman or CARS spectral images. The scheme works as a pre-selection to gather appropriate spectra which can be used as training data to establish a supervised classifier (e.g. Random Forest) to automatically identify subcellular compartments.
A dataset for testing can be downloaded here: http://www2.rz.rub.de:8234/imperia/md/content/pure/supplement.zip
Cite: Krauß, Sascha D., et al. "Colocalization of fluorescence and Raman microscopic images for the identification of subcellular compartments: a validation study." Analyst (2015). http://dx.doi.org/10.1039/C4AN02153C