A new tool for bioimaging based on super resolution Raman microscopy

Raman microscopy is a technique which interacts laser light of a particular wavelength with a target sample resulting in this light being scattered by the sample, the changes in energy of the scattered light is then measured. These changes in energy relate to vibrations from different molecules and produce a vibrational fingerprint of the sample relating to the molecular composition. When conducted using a microscope and a stage which moves, multiple Raman spectra in 2 and 3 dimensions can be acquired to produce an image of the sample based on the intensity and the location of particular vibrations within the sample. This is referred to as a Raman map and is very often a false colour map laid on top of a standard magnified microscope image of the sample, a white light image, e.g. a heat map of intensity of say a protein vibration overlaid on the image of a cell. Conventional Raman microscopy is normally in a confocal mode which means that the highest resolution in spatial terms is half the wavelength of the excitation light so typically around 250 nm. Biological structures and processes are on a much smaller scale and this is a limitation of Raman spectroscopy. An advantage of Raman spectroscopy is that it is label free and reliant on the specific molecular vibrations from the molecules in the interrogation volume, unlike fluorescence microscopy, which is the most commonly used form of optical microscopy in life sciences. However fluorescence microscopy requires addition of a label to the sample which changes the sample composition and can affect the intrinsic biological processes of a biological system. This proposal will produce a new tool to acquire Raman maps and then process the data to enhance the spatial resolution possible from a Raman confocal microscope. We propose to generate sub 100 nm spatial resolution using this tool which will greatly transform the use of Raman spectroscopy and microscopy in the life sciences. This tool will require no addition of labels or hardware modifications to existing Raman microscope instruments.

The tool proposed here involves the development of image processing algorithms to deconvolute data from intentionally oversampled Raman microscopy images of mesenchyemal stem cells. Typically, confocal Raman microscopy aims to work with a sharp focus and to generate clear images based on the molecular vibrations. However, the resolution in these images has historically been limited by the diffraction limit of light. Recent attempts to subvert this limit in Raman microscopy have been successful, but rely on the collection of a very large number (>100) of discrete Raman images, collected sequentially with a small spatial offset, to generate a super-resolved image. Given the large number of images required, the chance for sampling error, and spectral collection times needed in most biological samples, makes this strategy untenable. In our approach, we propose to address some of these challenges, bringing super-resolution to biological samples to identify and differentiate sub-cellular structure with molecular detail. We intend to oversample-collecting spectra at spatial intervals smaller than the diffraction limit of light- deconvolve and process the image, and reconstruct our super-resolution image from this oversampled data set. By combining sophisticated state of the art data algorithms with highly performing Raman instrumentation on a key biological sample, we will demonstrate the ability of our approach to go well beyond current, commonly obtainable spatial resolution using a Raman microscope. We see this as a key technological development to provide a simple and easy to use tool for use in the life sciences. We also envision the application of this tool to bioimaging to open new lines of investigation currently unimaginable due to technical limitations.

The main beneficiaries from this research will initially be academics as stated in the previous section. However, as the tool emerges in terms of its utility there will be initial beneficiaries arising from the commercial sector. There are opportunities for those involved in the manufacture of Raman instruments and we have a good relationship with a number of different instrument manufacturers such as Renishaw, Snowy Range, Cobalt and WiTech. There are also opportunities for those involved in software development and supply. Very often the instrument manufacturers use their own bespoke software which can limit users to specific instruments and data processing however we will consider whether our approach has patentability and as such explore various routes to maximum impact from this proposal. We see impacts going beyond the instrumentation and software in terms of the applications where we will have impact on the nation's health and wealth. This could be in terms of the assessment of new medicines, understanding more about regenerative medicine and its mechanisms for growth into viable three dimensional scaffolds. The aim of this proposal is to generate the proof of principal data for the tool and its benefits to academic and non-academic sectors with our routes to impact detailed in our pathways to impact statement.