Product Survey: High content screening systems

Obsessed with Details
by Harald Zähringer, Labtimes 03/2017

High content analysis systems are optimised for maximal light efficiency. But it’s worth nothing without intelligent data visualisation.

In the early days of microscopy, researchers who studied insects, small animals or cells had to draw little sketches to capture details of the magnified objects in their lab notebooks. Back then, to be a natural scientist not only required a keen mind, but also a sharp eye and solid drawing skills. Just think of the famous nerve cell illustrations of Spanish pathologist, Ramón y Cajal, that opened the door to modern neuroscience at the end of the nineteenth century.

Detailed drawing of astrocytes by Ramón y Cajal. Photo: Instituto Cajal

Nowadays, scientists still analyse microscopic photos by eye, utilising the remarkable ability of the human brain to extract and classify substantial picture data with very high precision. The experienced eye and brain of an expert pathologist, looking e.g. for morphological changes in microscopic slides of cancer cells, is still hard to beat even by sophisticated imaging analysis software.

The problem is that the throughput of human “picture analysers” is awfully slow. Hence, engineers have adapted classical wide-field or confocal fluorescence microscopes and turned them into High Content Screening (HCS) or analysis (HCA) systems. Though HCA is not necessarily associated with high throughput, HCA-systems are way faster then the eyes and brains of human analysers.

Four basic components

Most HCA-instruments contain four basic structural components, which are incorporated into rather unspectacular rectangular metal sheet housings: the optical elements of a wide-field or confocal fluorescence microscope, a scientific grade camera, a nanopositioning system that precisely moves the microscope stage holding the specimens in tiny steps, and an image analysis software.

The optical systems of wide-field HCA-instruments are very similar to classical inverted microscopes using mercury, xenon, and halogen lamps, or lasers and LEDs as light sources.

Confocal HCA-systems are slightly more modified. In standard confocal microscopes, light coming from a lamp, laser or LED falls through a fixed or adjustable pinhole onto the objective, thereby eliminating light which is out of the focus plane of the objective. By moving the sample stepwise through the light beam, the objective is scanned point-by-point. This linear scanning process, however, takes a lot of time and the intense single laser beam may bleach light-sensitive cell components.

Hence, confocal HCA-instruments are often equipped with a spinning or Nipkow disk that contains multiple small pinholes. While rotating with high speed, low intensity light passes the pinholes of the disk enabling short image acquisition times and reduced light exposure. Sometimes two coaxially aligned disks are combined. In this case, the pinholes of the upper disk contain microlenses that focus the light spots exactly on the pinholes of the lower disk. Due to the increased light throughput, lower laser intensities are necessary for imaging, which in turn reduces light exposure of the specimen.

To collect as much light coming from the irradiated sample as possible, objectives with high numerical aperture (N.A.) are mounted onto the nose pieces of HCA-systems. Though emersion objectives may complicate nanopositioning of probes, high N.A. oil or water objectives may also be installed in some instruments to further maximise light efficiency.

Probes are shifted automatically across the microscopic stage into proper position by extremely accurate nanopositioners. Finding the optimal focus plane of the probes along the z axis is pretty challenging and is usually done by laser or image-based systems. Laser-based techniques take advantage of light reflected at the bottom of the wells to automatically adjust the specimen into the focus plane. Image-based methods scan the specimen for special features, e.g., the contour of the nucleus, to compute the optimal focus plane with sophisticated algorithms.

Neither technique is perfect: laser-based autofocusing is fast but gets in trouble when the bottoms of the wells are not exactly flat. Image-based autofocusing on the other hand is considerably slower and may be fooled by artefacts or other structures not in the focus plane of the cells in the well.

Light emitted from the probes is directed to CCD or cMOS cameras and is subsequently visualised on a screen. Manufacturers of HCA-instruments follow two different camera philosophies, either preferring CCD- or cMOS cameras. Thermo Fisher, for example, belongs to the CCD-camera camp, arguing that CCDs show higher quantum yields compared to cMOS-chips, as well as better signal-to-noise ratios. Other companies, like Perkin Elmer, favour cMOS cameras, or offer both types in their instruments, such as Molecular Devices.

Stringent analysis pipeline

But independent of the installed camera, the trickiest part of high content screening starts after the photo shooting with the bias-free analysis and processing of image data. Both commercial and open source HCA-image analysis programs follow a simple and straightforward analysis pipeline to extract relevant picture data.

In step one, artefacts, illumination flaws, and high background noises are removed. In step two, the software chooses appropriate segments of the image that are further processed in step three, in which objects suitable for analysis are selected.

Following this pre-selection procedure, the software extracts relevant cell features and visualises the resulting datasets, for example, as heat maps, line graphs, histograms, scatterplots or network graphs.

However, making sense out of abstract line graph or heat map datasets is not very intuitive. The visualisation only represents an indirect “image” of the cells – you won’t see what they actually look like under the microscope. Hence, researchers have developed glyph-based visualisation programs, to visualise cellular features in a sketch that better resembles a cell. An image is worth a thousand words – this phrase is still valid more than 100 years after Ramón y Cajal’s seminal nerve cell drawings.

First published in Labtimes 03/2017. We give no guarantee and assume no liability for article and PDF-download.

Table of Products as PDF-download: pdficon Formatted for Printout

Information 4

Information 5

Information 6