CID-Net (Complex-valued Incection for Diverging-wave Network)

This repo provides CID-Net model with trained weights in Open Neural Network Exchange (ONNX) format and a inference demo script, for the following paper:

Complex Convolutional Neural Networks for Ultrafast Ultrasound Imaging Reconstruction from In-Phase/Quadrature Signal

Repo Contents

• model/cidnet.onnx: CID-Net ONNX model with trained weights

• data/

  • grid_x.h5: image grid in lateral direction
  • grid_z.h5: image grid in axial direction
  • data_gammex.h5: data example acquiered from phantom (Gammex, model 410SCG)
  • data_cirs.h5: data example acquiered from phantom (CIRS, model 054GS)

• demo.ipynb: Running inference

Requirements

• numpy
• h5py
• onnxruntime
• matplotlib

Model inference with the image examples

Code pieces are provided in jupyter notebook demo.ipynb