A Sparse-View CT Reconstruction Method Based on Combination of DenseNet and Deconvolution.

The plan is to try the DD-Net on our sparse-view data. The original paper: https://ieeexplore.ieee.org/document/8331861 The original project GitHub page: https://github.com/zzc623/DD_Net

What is deconvolution?

Convolution is when we take the image and try to shrink it. Deconvolution is basically the same as a convolution layer, except instead of subsampling we do upsampling and the ‘stride’ is interpreted oppositely. So stride=2 would do upsampling by 2 (output has twice more things) instead of downsampling by 2. Example below is for deconvolution in python.

u6 = Conv2DTranspose(64, (2, 2), strides=(2, 2), padding='same') (c51)