https://stanfordmlgroup.github.io/projects/mrnet/

1. The Final Classification Networks

class Classify[source]

Classify(in_dim, mid_dim, out_dim, p_drop=0.5) :: Module

The Classifier on top of the triplenet network

Proposed extension: Maintaining the spatial Relation using Recurrence

class ClassifyRNN[source]

ClassifyRNN(device, input_size, hidden_size, n_layers=1) :: Module

The Classifier on top of the triplenet network

2. The Complete Network with Backbone and Pooling

class Compressor[source]

Compressor(device, args, train_data=None, backbone='resnet18', training=True, document=True) :: Module

This class compresses the data from all slices to be only a vector. Contains the backbone and the pooling

3. The TripleNet with Losses and bringing it all together

class TripleMRNet[source]

TripleMRNet(device, args, train_data=None, backbone='resnet18', training=True, document=True) :: AbsModel

adapted from https://github.com/yashbhalgat/MRNet-Competition with the knowledge of: https://journals.plos.org/plosmedicine/article?id=10.1371/journal.pmed.1002699