Gen7Net is a Convolutional Neural Network for classifying prostate cancer in MRI images. It was built with TensorFlow and Keras and is based on MobileNetV2. Training images were taken from the Cancer Imaging Archive.
Dependency handling was performed with nix-shell for this project. Make sure to have nix installed. Enter the dev environment with:
nix-shell shell.nixA bash script for converting .dcm to .jpg.
Run with:
cv.sh <command> <source> <destination>- help | h: display help message
- convert | cv: convert images from
.dcmto.jpg - clean | clear | clr: remove all
.jpgfiles from <source> - transfer | copy | extract: transfer converted files to <destination> using rsync
Dependencies are handled by running nix-shell against shell.nix.
- detox: rename files and folders with unicode standard
- findutils: find any
.dcmor.jpgfiles in the source folder - rsync: transfer large quantities of files from A to B
- tree: displays any folder structure in a tree view
- imagemagick: does the heavy lifting of converting between image formats
jupyter-notebook for Gen7Net. I recommend running it with Google Colab. In order to run it locally, I recommend using docker containers with a tensorflow packaging tool like tensorman.
- tensorflow
- keras
- mobilenetv2
- pandas
- numpy
- jupyter-notebook