Untrained CNN and coefficient-space regularization for joint reconstruction in photoacoustic tomography

Key points

• A regularized data-driven framework is proposed in which the unknown fields are represented in finite-dimensional wavelet spaces and fixed nonlinear features generated by an untrained convolutional neural network (CNN) are used to reconstruct the initial-pressure coefficients through Tikhonov-regularized least squares.1
• A learned coefficient-space prior is subsequently used to infer the squared sound-speed coefficients.1
• In comparison with the supervised trained-CNN approach considered in this work, the untrained representation yields substantially smaller reconstruction errors while requiring approximately one-sixth of the computational time.1
• These results demonstrate that fixed random convolutional features combined with regularized coefficient-space learning provide an efficient and robust approach to joint reconstruction in photoacoustic tomography.1

Sources (1)

• [1]Untrained CNN and coefficient-space regularization for joint reconstruction in photoacoustic tomography
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 16, 11:20 AM

  The joint reconstruction of the initial pressure and spatially varying sound speed in photoacoustic tomography from boundary measurements is considered.

  A regularized data-driven framework is proposed in which the unknown fields are represented in finite-dimensional wavelet spaces and fixed nonlinear features generated by an untrained convolutional neural network (CNN) are used to reconstruct the initial-pressure coefficients through Tikhonov-regularized least squares.

Extractive summary: sentences quoted from the sources.