SegmentationTools

QMRITools package: functions for segmenting muscle data using convolutional neural networks.

MakeUnet — Makes a 2D or 3D UNET.

MakeNode — Makes a node for a UNET.

SegmentData — Segments out of phase Dixon muscle data.

$debugUnet — If set True the MakeUnet and MakeNode can be debugged.

AddLossLayer  ▪  DiceLossLayer  ▪  FocalLossLayer  ▪  JaccardLossLayer  ▪  TverskyLossLayer

DiceSimilarity  ▪  SurfaceDistance  ▪  JaccardSimilarity  ▪  MakeDistanceMap

MakeChannelClassImage  ▪  MakeChannelImage  ▪  MakeClassImage  ▪  MakeChannelClassGrid

ImportITKLabels  ▪  MuscleLabelToName  ▪  MuscleNameToLabel

ClassifyData  ▪  MakeClassifyImage  ▪  MakeClassifyNetwork

ClassDecoder  ▪  ClassEncoder  ▪  DataToPatches  ▪  PatchesToData

TrainSegmentationNetwork  ▪  AugmentTrainingData  ▪  AugmentImageData  ▪  PrepareTrainingData  ▪  GetTrainData  ▪  CheckSegmentation  ▪  ShowTrainLog

ApplySegmentationNetwork  ▪  SplitDataForSegementation  ▪  SegmentDataGUI  ▪  GetNeuralNet  ▪  NetDimensions  ▪  ChangeNetDimensions  ▪  AnalyseNetworkFeatures  ▪  NetSummary