insam/deepmiil

3D-volume Conv Pix2Pix cGans for PET/MR

DeepMIIL

3D-volume Conv Pix2Pix cGans for PET/MR

Based on pix2pix by Isola et al. Directed conversion of affinelayer's tensorflow implementation. https://github.com/affinelayer/pix2pix-tensorflow

To run -Input folder: String that points to the directory containing the 3D volumes. Each volume should be represented by its own sub-folder containin all the individual slices. Each slice of the volume should be named numerically based on their order. [IE: 1.png, 2.png, 3.png, ..., xxx.png]

Autov: MSE and MS-SIM accuracy benchmarks.

autov compile (temp makefile replacement): g++ autov.c++ -o autov pkg-config --cflags opencv pkg-config --cflags --libs opencv