Output details

Parallelization methods for implementation of a magnetic induction tomography forward model in symmetric multiprocessor systems

<28> This paper presents key findings from an EPSRC funded project. It is the most comprehensive work addressing the computational challenge of real time magnetic induction tomography (MIT) which hinders the use of this imaging technique in medical and industrial applications, such as stroke detection and pipeline oil/water monitoring. A dual distributed and shared memory approach is proposed, which uses parallelisation of coil and physical domain loops. It has been shown to yield fast image reconstruction paving the way for practical low cost and portable MIT monitoring devices. This work attracted interest which initiated a collaboration with Philips Research Laboratory.