Accelerating 4D PhotoAcoustic Tomography

Monday, April 29, 2019 - 9:00am - 9:50am
Lind 305
Simon Arridge (University College London)
PhotoAcoustic Tomography (PAT) has become established as a significant imaging modality allowing the qualitative imaging in 3D of absorbed optical energy in the visible and near-infrared range, with high resolution, by exploiting the conversion of optical to acoustic energy and the discontinuity preserving propagation of ultrasound under the assumptions of the wave equation. Similar to Magnetic Resonance Imaging (MRI), the high resolution achieved carries with it the cost of relatively long acquisition times. In this talk I will discuss our work over several years to accelerate the data acquisition step using subsampling. Inspired by success in MRI we developed compressesed sensing methods for both 3D and 4D (i.e. 3D + time - dynamic imaging), as well as parallel measurement instrumentation. More recently we aslo developed deep learning methods both for the reconstruction step and for fast forward modelling using approximate models and learned corrections thereof. Results from simulations and experiments have brought real time (video rate) PAT within reach.
Joint Work with : P.Beard, M.Betcke, B.Cox, A. Hauptman, N.Huynh, F.Lucka,