Learning based sound speed estimation and aberration correction in linear-array photoacoustic imaging

Mengjie Shi, Tom Vercauteren, Wenfeng Xia

Paper:https://arxiv.org/abs/2306.11034

Implementation a fully convoluntional neural network for sound speed estimation. The network was dedicated to an input of (128, 1024, 1) acquired from a plane wave transmission with a linear array Ultrasound (US) probe. The network was pretrained with a large number of pure simulated data and fine-tuned with a small amount of phantom data.

test_data.mat can be used to test the network. Trained models were saved in saved_models folder.

Datasets：

Simulated US (used for model training): https://drive.google.com/file/d/1lydLrlmynVdLWVfiAmf_8J01J6gU1cvR/view?usp=drive_link

US simulation script: https://drive.google.com/drive/folders/1oyiXrN_xsvKDdkKyKniOvYuwiFBdUXj6?usp=drive_link

Dependencies: k-Wave, ustb

Please cite the paper if you tend to use the codes, dataset, or find it helpful for your work. Contact: mengjie.shi@kcl.ac.uk