High-Content Microscopy & Data Mining
Concept of automated correlative microscopy: Interesting events/targets (hits) are localised during high-throughput screens and further analysed using more sensitive detection methods.
In fluorescence microscopy three major techniques are commonly used:
- wide field microscopy
- PROS: robust, fast, high-throughput capable
- CONS: low information content, 2D resolution
- confocal microscopy
- PROS: high information content, 3D resolution
- CONS: only medium-throughput capable
- super-resolution microscopy
- PROS: single molecule resolution
- CONS: demand on dyes and substrates, no automation (so far)
Thus, a combination of these techniques would result in an integrated screening platform for
high-throughput → high-resolution → super-resolution
imaging. Within such a correlative microscopy platform, high-throughput (wide field) data acquisition is considered as a trigger / filter system which detects interesting events (hits) on-line during an automated large scale screen (1). The selected candidates can then be analysed more detailed downstream of the microscope pipeline via an automated switch to downscaled high-resolution confocal screening (2) which again triggers for hits in high-resolution detection mode. The overall screening pipeline thus ends up in super-resolution imaging (3) of highly interesting - not only rare - events resulting in a maximum of information content for any biological assay.
(1) order of magnitude of data sets: 106
(2) order of magnitude of data sets: 104
(3) order of magnitude of data sets: 102