Acoustic Tomography

Acoustic tomography aims at recovering the parameters of an unknown medium by studying the effects of the medium on the sound propagating through it. This requires first an accurate forward model which describes well the underlying physical system and second, measurements of good quality. Then, the unknown model is estimated by solving a nonlinear inverse problem which looks for a model that describes well the measurements. The accuracy of the forward model, the fidelity of the measurements and the choice of the inversion method all have direct influence on the quality of the estimated model. This forms the basis of acoustic-based imaging systems, a powerful imaging modality which has been successfully applied in medicine, industrial non-destructive testing, seismology and oil exploration.

Our research on inverse problems and tomography includes:

  • Dictionary learning for sparse representation of medical images
  • Ray-based and wave-based solutions to inverse problems in acoustic tomography
  • Sparsity-based regularization techniques for inverse problems in acoustic tomography
  • Sampling requirements in tomographic imaging
  • Travel-time estimation for bent-ray acoustic tomography
  • Fast and efficient implementation of wave propagation using GPU/FPGA
  • Tomographic reconstruction from stochastic measurements
  • Tomographic sensing and reconstruction in epidemiology

Current Research Projects

  • Learning Dictionaries and Inverse Problems

 

Past research projects

  • Tomographic field reconstruction from stochastic measurements
  • Tomographic field reconstruction using a mobile sensor network
  • Convex optimization for X-ray CT
  • Acoustic tomography based on wave equation
  • Compressed sensing with probabilistic measurements for solving epidemic problems
     

People

Ali Hormati, Ivana Jovanovic, Reza Parhizkar, Martin Vetterli

 

WWW.USENSE.ORG

We have developed USense.org, a platform designed to freely distribute open source software, hardware and data for ultrasound medical applications. The main goal of this website is to promote the  reproducibility of medical ultrasound research by providing users with the complete package needed to reproduce results presented in scientific publications. This typically includes documented software code, description of hardware design and ultrasound data. The website now contains the following codes available to public:

  • WaveTomography: is a 2D time-domain waveform tomography reconstruction algorithm that can be run on graphical processing units (GPU).
  • Mavilib: ///  is a very simple C++ library that allows to transfer data between your C++ program and a Matlab workspace, as well as to invoke Matlab commands from your C++ application.
  • WaveletLib: ///   provides a set of C++ functions to compute multi-level wavelet transforms of one-dimensional and two-dimensional signals using the lifting scheme.