brain-predict

Valið efni í rafmagns- og tölvuverkfræði (RAF609M vor 2019)

Description:

This is an implementation of a 3D-CNN network used to predict sex and age from T1w MRIs

File:

Conv3D.ipynb        - Jupyter notebook with CNN implemention  
plot_results.ipynb  - Jupyter notebook for visualizing results  
parameter_search.py - Performs hyper-parameter with Hyperopt  
read_results.py     - Shell script for reading .csv with parameters and results from parameter_search.py  
utils.py            - Various functions used in the code  
conv3D-weights.h5   - Weights for a pre-trained network  
dropout30epochs_results.csv - Results and training data from parameter search over 30 epochs

Folder:

data                - Contains preprocessed data and labels  
data_preprocess     - Contains various code used to preprocess the data