Conference Agenda

Miles Padgett and Simon Mekhail, The University of Glasgow

Endoscopic imaging systems based upon bundles of optical fibres are commonplace across medical and industrial applications. However, even just one of these optical fibres, less that 100µm in diameter, transmits enough spatial modes to relay an entire image, but intermodal dispersion rephases the output modes such that any input image becomes unrecognisable at the output. This problem can be overcome by measuring the transmission matrix of the fibre and using the inverse of the matrix to set the required input light fields to produce a scanning spot over the scene at the output. The backscattered light from this spot can then be measured to create an image of the scene. A current limitation is that the matrix is time consuming to measure and the required input beams time-consuming to calculate. Furthermore, each time the fibre is moved the matrix needs to be remeasured and the input beam recalculated. Here we show that the use of high-speed cameras and GPU computing can reduce this measurement and calculation time to a few 10s seconds, inspiring new modes of operation. When combined with the use of graded index fibres to minimises the sensitivity of the matrix to movement of the fibre, we show that 5-10 rapidly pre-recorded matrices are sufficient to create an imaging system that works over a wide range of fibre positions, giving near continuous imaging from a compact instrument. Such ultra-minimally invasive imaging systems have many uses in inspection and medical applications.