Endoscopy

Initially we implemented wide-field FLIM endoscopy with a rigid arthroscope at up to 29 fps [1] and we also implemented wide-field FLIM with flexible endoscopes incorporating fibre-optic bundles [2]. Initially we used frequency-doubled mode-locked Ti:Sapphire lasers to provide the ultrashort pulsed excitation for FLIM but subsequently we demonstrated wide-field FLIM endoscopy with a flexible optical fibre bundle endoscope using blue gain-switched diodes to provide picosecond excitation pulses [3].

For in vivo application, there is a trade-off between minimising light dose to the tissue (to avoid any phototoxicity) and acquiring sufficient photons to fit the fluorescence decay profiles to complex decay models. This can be addressed by binning all the image pixels or by using global fitting techniques. For the former approach, the instrumentation can be simplified if the goal is endoscopic time-resolved spectroscopy rather than imaging and we addressed this by developing a series of instruments utilising compact fibre-optic bundles incorporating excitation and detection fibres. These instruments are multidimensional fluorometers providing single-point measurements. The first such instrument utilised a compact frequency-tripled ultrafast Yb-doped glass fibre laser providing picosecond pulses at 355 nm and gain-switched diode laser providing picosecond pulses at 445 nm. This instrument was evaluated in vivo to diagnose skin cancer [4]. This instrument was also applied to explore studies of autofluorescence lifetime as a potential diagnostic readout for osteoarthritis through ex vivo studies of cartilage [5]. Subsequently a more compact version of this instrument utilising gain switch diode lasers at 375 nm and 435 nm was applied to ex vivo [6] and in vivo studies exploring autofluorescence lifetime as a readout of neoplasia in the human colon. A further version of this instrument that also incorporated while-light reflectance measurements was applied to in vivo preclinical studies of heart disease in a rat model [7] and to ex vivo studies of a Langendorff isolated-perfused rat heart model [8].

We also translated laser scanning confocal FLIM to endoscopy, developing a confocal FLIM endomicroscope based on the Mauna Kea Technologies CellVizio platform, to which we added a frequency-doubled mode-locked Ti:Sapphire laser to provide pulsed excitation and TCSPC detection [9]. We demonstrated that this provided comparable performance to our lasers canning confocal FLIM microscope in terms of lifetime determination and was able to realise confocal FLIM endomicroscopy of fluorescent-protein labelled cells at up to 10 fps. This instrument was applied in a preclinical study with colleagues at the Francis Crick Institute to quantify drug-target engagement in vivo in a murine cancer model utilising FLIM of FRET between the fluorescent anticancer (DNA-intercalating) drug, doxorubicin, and EGFP, which was expressed in the mouse histones [10].

We were motivated to apply FLIM endoscopy to study autofluorescence in vivo for label-free studies of disease but the required excitation wavelengths also excite significant unwanted luminescence in the glass from which the optical fibre bundles in the endoscopes are made. This results in a problematic background that compromises FLIM data and reduces the dynamic range of the endoscope. Multiphoton microscopy avoids this issue through the longer excitation wavelengths used but instead can be compromised by the pulse broadening that results from self-phase modulation and group velocity dispersion in the optical fibre bundles. This can be minimised by reducing the peak intensity of the excitation pulses and to this end in 2008 we invented and patented the concept of spreading the excitation pulse energy over multiple cores of an optical fibre bundle optic bundle (or multiple modes in a multimode optical fibre) and utilised a proximal spatial light modulator (SLM) to control the relative phase of the distal light emerging from different cores or modes. We demonstrated that controlling the phase of the light from different cores enabled scanning and focussing of the distal light with no distal optical or scanning components [11]. Working with the University of Bath, we subsequently showed that we could enable multiphoton endoscopic imaging with dynamic correction of the phase control at up to 100 Hz [12] that enabled us to image through an optical fibre bundle undergoing motion that perturbs the relative phases in the fibre cores. We then showed that we could dynamically control the relative phases of the light from the fibre cores at >400 Hz using only proximal measurements [13], which is a key step towards ultracompact multiphoton endoscopes. We also developed a semi-random multicore fibre design that allowed a larger field of view to be imaged, and we also showed that multiphoton excitation allows the number of independent image resolution elements achieved to exceed the number of fibre cores [14].