Deep Neural-Network Based Sinogram Super-resolution and Bandwidth Enhancement for Limited Data Photoacoustic Tomography
This MATLAB code was used as part of the work presented in
Navchetan Awasthi*, Gaurav Jain*,Sandeep Kumar Kalva, Manojit Pramanik, Phaneendra K. Yalavarthy, “Deep Neural-Network Based Sinogram Super-resolution and Bandwidth Enhancement for Limited Data Photoacoustic Tomography,” in IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control (Special issue on Deep Learning in Medical Ultrasound) 2020 (DOI: 10.1109/TUFFC.2020.2977210).
- Co-first authors with equal contribution **The raw measurement data for the experimental experiments is not provided and can be requested. ***Please contact if you find any mistakes or if you need any help regarding the codes.
The data for the above work can be obtained by using the google drive link and requesting for access: https://drive.google.com/drive/u/3/folders/12egdPGu3muwBCyhYapntNpbAUTFDEJm_
#Matlab implementation for testing the phantoms: Testing_script.m #Matlab implementation for adding noise : addnoise.m #Matlab implementation for generating the patches : generate_patches.m #Matlab implementation for generating the System Matrix : sysBuildPAT_mod_Band.m #Matlab implementation for generating the test data : test_data_generation.m
#Python implementation for loading the training and validation data : load_train_val_data.py #Python implementation for the loss function for our network : loss.py #Python implementation for testing the network : test.py #Python implementation for SRCNN network: srcnn_model.py #Python implementation for training SRCNN network: srcnn_train.py #Python implementation for U-Net-Relu network: unet_relu.py #Python implementation for training the U-Net-Relu network: train_unet_relu.py #Python implementation for U-Net-Elu network: unet_elu_All.py #Python implementation for training the U-Net-Elu network: train-unet-elu-all.py #Python implementation for the U-Net-Hybrid architecture : unet1.py #Python implementation for training the U-Net-Hybrid network : train.py