Imaging in the life sciences tends to be characterised by a lack of light. Shine too much light on a cell and it dies, which is why biologists use illumination sparingly. The downside to the super resolution microscopy techniques, which won the Nobel Prize in Chemistry in 2014 and manage to beat the diffraction limit of light by some clever fluorescence imaging methods, is that they rely on quite high laser intensities – not great for live cell imaging.
Now, other techniques like light sheet microscopy, also known as selective plane illumination microscope (SPIM), are being developed, which use much lower light intensities and are ideal for live cell imaging, but demand more from the image sensors.
At Photonics West, which will take place in San Francisco in February 2016, Dr Ryan McGorty, assistant professor at the University of San Diego, will give a presentation on a SPIM system he and his colleagues developed at the University of California, San Francisco (UCSF) for high content biological imaging. The system McGorty built is designed to take SPIM and make it compatible with high throughput imaging platforms, like 96-well plates or microfluidic devices.
‘To date, light sheet microscopes have been made to make the sample accessible to the optics,’ McGorty commented. ‘However, that’s not compatible with sample mounting procedures found in other more established microscopy techniques.’
In most light sheet microscopes the sample is placed between two objectives, one creating the sheet of excitation light and the other imaging that focal plane. The sample can be viewed from many different angles by rotating it through the light sheet. However, because the sample mounting procedure is so specialised, it’s hard to do high throughput imaging.
The open top SPIM system developed at UCSF would help researchers who want to, for instance, screen for a library of chemicals and see how those chemicals affect embryo development. To do this, the researcher might observe a number of embryos each given a different treatment. ‘If you wanted to do that kind of screen on whole organisms, ideally you’d use some sort of optical sectioning technique that isn’t too toxic to the organism,’ explained McGorty. ‘That’s why light sheet microscopy in general would be beneficial, and our microscope would be particularly beneficial in that it would allow that kind of light sheet technology to be incorporated into high-throughput screening.’
The technique acquires a stack of images as a 3D data set by moving the sample through the light sheet with a motorised stage. The group is currently working on using adaptive optics to better correct aberrations in the system, in order to work with high numerical aperture objectives and obtain higher resolutions.
Scientific CMOS
To capture the stack of images from a light sheet microscope requires a fast and sensitive image sensor. McGorty’s open top SPIM system uses a scientific CMOS (sCMOS) camera. Dr Gerhard Holst, head of research at camera company PCO, commented that super resolution microscopy in general can require frame rates up to 1,000fps, which is achievable at reduced resolution with sCMOS sensors.
PCO sells sCMOS cameras to manufacturers of SPIM, structured illumination microscopy (SIM) systems, and super resolution microscopes, including Carl Zeiss for its light sheet microscope and GE for its SPIM systems.
Scientific CMOS was developed by Fairchild Imaging, now part of BAE Systems, PCO, and Andor Technology, now owned by Oxford Instruments. The sCMOS project started in 2006/2007 with Fairchild Imaging suggesting new structural improvements to CMOS sensors. Holst commented: ‘At the time, the general opinion was that if you want high frame rates you go for CMOS image sensors, whereas if you want good image quality and a sensitive low readout noise signal then cooled CCDs are the way to go.’
Fairchild Imaging, PCO and Andor combined forces to develop a new type of CMOS image sensor providing low readout noise, a sufficiently high frame rate at high resolution, and a high dynamic range. The sensor produced a much higher frame rate compared to cooled CCDs.
Scientific CMOS is a standard CMOS sensor made in the same fabs as other CMOS wafers. The improvement in performance comes from the pixel architecture, explained Holst. ‘Scientific CMOS has become very popular and replaced lots of the electron multiplication CCD devices,’ he said. ‘Since sCMOS is sensitive, has an extremely low readout noise (one electron), and can image at up to 100 frames per second, it became very appealing for high throughput life science applications.’
Life science applications are largely based on the measurement of fluorescence using biomarkers, which – in most cases – doesn’t have high quantum efficiency and therefore requires sensitive cameras. Laser diodes tend to be used as a light source. ‘Being able to get low-cost, relatively high-power light focused onto a small area is very useful, and that’s something lasers excel at above any other lighting technique,’ explained James Saxon, a technical sales engineer at photonics company Laser Components.
Laser Components sells laser diodes to customers making bioinstrumentation. ‘High light intensity gives high contrast images. However, too much light could damage the sample,’ noted Saxon. ‘There is normally a trade-off between acquiring the contrast necessary for image analysis while at the same time not damaging the sample.’ High contrast imaging would typically be used for cell counting, for instance.
The cameras also have to be stable because, in many cases, when imaging multi-well plates or other high-throughput methods, the scientists want to compare images. ‘It’s an interesting time for using sCMOS for life sciences applications,’ commented Holst. He said that more methods are emerging from companies supplying high-throughput biological imaging.
One example is digital pathology using slide scanners to scan multiple biopsy samples. These systems typically use CCD-based colour cameras for colour measurements and sCMOS cameras for fluorescence measurements. The slides are moved automatically and software digitally stitches the images together. ‘We [PCO] enjoy a number of these applications and sell 500 to 800 cameras per year to OEMs making instruments for high throughput imaging in life science,’ Holst said.
Fairchild has recently improved the process of making sCMOS wafers and has achieved 10 per cent higher quantum efficiency – the global shutter version now has 70 per cent QE while the rolling shutter sensor has 80 per cent QE. Andor has recently released its Zyla 4.2 Plus camera based on the latest generation sCMOS sensor, offering 82 per cent QE. Hamamatsu also supplies sCMOS cameras.
‘Scientific CMOS still has some weak points,’ acknowledged Holst. ‘The noise behaviour is not Gaussian; there is a significant “noise tail”, for instance. So there is ongoing work on improving this and reducing the number of noisy pixels.’
Bio-machinery
Cellular imaging can be highly specialised. An embryo, for instance, is nearly transparent and its structure is only revealed with sophisticated imaging techniques. ‘To the naked eye, and without the assistance of some fairly advanced imaging techniques, many cells or tissue samples can appear to be invisible,’ commented Martin Price, a director at Munich-based Opto, a company building optomechatronic systems. ‘Add to that the fact that such transparent cells may be suspended in a transparent liquid medium, and in a transparent well plate or petri dish, and you have a real imaging challenge to contend with.
‘Historically, in order to do that you would need a fairly well-specified inverted microscope, configured to deliver some advanced microscopy techniques, such as phase contrast and multi wavelength fluorescence,’ Price continued. ‘Techniques such as phase contrast enhancement are used to image the structure of an otherwise transparent cell by converting phase shifts in light passing through a transparent specimen and other objects to contrast changes in the image – making the invisible, visible.’
Large benchtop microscopes employ classic techniques like fluorescence, polarisation and contrast enhancement to image cells, but often the number of features available on these systems make them complex and expensive, and a challenge to integrate into a machine.
‘To the machine builder, there can be a huge amount of unnecessary cost and complexity attached to integrating a classical microscope, which drove us to develop embedded modules,’ explained Price.
Many of the solutions Opto builds are based on simple modular elements that still offer the techniques found in a full benchtop microscope, but which are condensed down into small, low-cost, easy to integrate modules.
‘Engineers involved in the design of the latest automated cellular engineering machinery frequently need the raw imaging performance of the best microscopes, but delivered in an ultra small footprint, high performance industrialised module designed exactly to the needs of their machine,’ Price said.
The modules that Opto builds differ depending on whether the instrument is a high-end system or an inexpensive handheld device. ‘A low-end handheld device may use lightweight moulded plastic optics; you would be looking to extract the maximum performance out of something that was very inexpensive and mass produced. It may use low-cost CMOS technology, moulded optics, packaged in a small form factor to meet the cost target,’ Price explained.
‘As instrumentation has become more and more complex, we have begun to see a distinct shift away from clients making do with off-the-shelf components and moving towards the cost / performance benefits and long-term efficiencies of a volume custom solution, and that’s precisely where we fit in,’ he continued. ‘Frequently, we can build a completely customised module in small to medium volumes for a lower cost than building a system out of standard off-the-shelf components.’
Price commented that Opto’s largest growth area is in developing imaging modules for integration into time lapse cellular imaging machines and other fully automated, and remotely located cellular engineering instruments.
‘The case for automated cellular diagnostics is significant,’ Price remarked. In the past, laboratories have relied on lab technicians performing manual cell evaluations, whereas now a lot of these processes are automated. Now, sCMOS technology, and clever software algorithms, along with the kinds of embedded imaging modules built by Opto, are enabling high-throughput and automated cellular imaging. Price said that ‘with the added potential of internet connectivity and associated remote diagnostics, the possibilities remain significant.’
