Smart acquisition and data reduction for light-sheet microscopy of the cell cycle

New developments in light microscopy are making possible previously impossible experiments, from super-resolution to high-speed imaging. One type of state-of-the-art microscope, the light-sheet microscope, is revolutionising cell biology by enabling very high speed three-dimensional imaging while projecting substantially less laser light into the sample. The result is a far gentler microscope than traditional laser-based microscopes and as a result cells can be observed for far longer without suffering phototoxic damage. It works by focusing a plane of light - a light sheet - onto the sample and then using a second lens to detect the emitted light from fluorescent molecules. Due to the high-speed image capture rate and the vastly lengthened safe imaging potential, light-sheet microscopes are set to produce truly "big data" with multiple terabytes per experiment.

This project aims to develop new "smart" software to enable light-sheet microscopes to analyse the samples of cells being imaging and independently make decisions to adjust the imaging protocol. For example, the microscope will be able to focus down on cells only to avoid wasting space on the background and then among those cells select only those in mitosis - the process of cell division. The acquired data will then be quantitatively analysed for the motion of particles and to take fluorescence measurements to allow researchers immediate access to results while also automatically being subjected to data reduction to reduce the big data burden.

We aim to develop smart light-sheet microscope control software and data reduction and analysis software to aid cell cycle researchers in taking full advantage of these state-of-the-art technologies. Modern light-sheet microscopes are set to revolutionise cell biology by offering high-speed imaging with minimal phototoxicity, enabling timelapse experiments to be performed for far longer than previously possible and with greater detail. These capabilities, however, present new challenges for the researcher in storing, managing and analysing extremely large datasets on the terabyte scale when using the full capabilities of the technology.

To alleviate some of these challenges for cell biologists, we will develop a real-time image analysis system to segment cells and classify their cell cycle and mitotic phases using machine learning. Based on this analysis our software will instruct the microscope to restrict or modify its acquisition protocol, limiting capture to only cells fitting requirements specified by the user. Following image capture, we will introduce a pipeline that will automate the data reduction process, depending on the nature of the experiment, to include downsizing, downsampling and compressing data. Further to this we will integrate an extendable analysis module supporting particle tracking and fluorescence quantification, to enable immediate access to quantitative results. This will, in many cases, allow storage of reduced data only relieving the burden of excessive data storage requirements.

Microscopy users
Scientists using light-sheet microscopes will be beneficiaries of the project. Our software will provide tools to reduce the burden of big data in state-of-the-art light-sheet microscopy experiments and widen the technology to a broader range of scientists, many of whom are not specialists in large-scale data analysis and reduction.

Microscopy developers
Commercial organisations developing microscopy systems and software will stand to benefit from the project. We will make the software available as open-source, hence others can use and develop the software further, with appropriate attribution. In particular, we are working closely with Intelligent Imaging Innovations (3i) Ltd. (see attached Letter of Support) and will seek to integrate modules into their system where possible. Therefore, there is a direct potential economic benefit and through this integration we will achieve a wider dissemination of our software contribution.

Public health
Our software for making light-sheet microscopes smarter and easier to use is targeted toward cell cycle researchers. Correct functioning of the cell cycle is crucial for health and problems are implicated in numerous health problems, including cancer, reproduction and aging. Advancements which enable previously prohibitively difficult experiments in these areas may lead to direct benefits to the health and well-being of the population of the UK and the world. Furthermore, the cancer drug market is worth over $100bn globally (Forbes, 2015) and potential future drug development linked to the outcomes of this research will positively impact the economic status of the UK.

Public outreach
Scientists involved in outreach programs to try to inspire school children to opt for STEM (science, technology, engineering and mathematics) careers will benefit from this technology. The visual nature of high-speed multicolour timelapse microscopy of cells and the intelligent automatic control of the system are ideal for captivating the interest of the public.

Academic impact
The approach taken in this project will also have an impact in the academic environment by providing an example of research bridging disciplines and cultures between computer science and cell biology. The co-I will have gained further skills in cell culture, microscopy and software development, in addition to academic writing and presentation skills. After completion, the staff will use their skills to contribute to academia, in the host lab or elsewhere.