class is a template class for using with GUI developed using appdesigner of Matlab More...

Inheritance diagram for mibDeepController:

Collaboration diagram for mibDeepController:

Public Member Functions
function [ lgraph , outputPatchSize ] =	createNetwork (previewSwitch)
	generate network Parameters: previewSwitch: logical switch, when 1 - the generated network is only for preview, i.e. weights of classes won't be calculated

function net =	generateDeepLabV3Network (imageSize, numClasses, targetNetwork)
	generate DeepLab v3+ convolutional neural network for semantic image segmentation of 2D RGB images

function net =	generate3DDeepLabV3Network (imageSize, numClasses, downsamplingFactor, targetNetwork)
	generate a hybrid 2.5D DeepLabv3+ convolutional neural network for semantic image segmentation. The training data should be a small substack of 3,5,7 etc slices, where only the middle slice is segmented. As the update, a 2 blocks of 3D convolutions were added before the standard DLv3

function [ net , outputSize ] =	generateUnet2DwithEncoder (imageSize, encoderNetwork)
	generate Unet convolutional neural network for semantic image segmentation of 2D RGB images using a specified encoder

function [ patchOut , info , augList , augPars ] =	mibDeepAugmentAndCrop3dPatchMultiGPU (info, inputPatchSize, outputPatchSize, mode, options)

function [ patchOut , info , augList , augPars ] =	mibDeepAugmentAndCrop2dPatchMultiGPU (info, inputPatchSize, outputPatchSize, mode, options)

function TrainingOptions =	preprareTrainingOptionsInstances (valDS)
	prepare trainig options for training of the instance segmentation network

	mibDeepController (mibModel, varargin)

function	closeWindow ()
	update preferences structure

function	dnd_files_callback (hWidget, dragIn)
	drag and drop config name to obj.View.handles.NetworkPanel to load it

function	updateWidgets ()
	update widgets of this window

function	activationLayerChangeCallback ()
	callback for modification of the Activation Layer dropdown

function	setSegmentationLayer ()
	callback for modification of the Segmentation Layer dropdown

function	toggleAugmentations ()
	callback for press of the T_augmentation checkbox

function	updateBatchOptFromGUI (event)

function	singleModelTrainingFileValueChanged (event)
	callback for press of SingleModelTrainingFile

function	selectWorkflow (event)
	select deep learning workflow to perform

function	selectArchitecture (event)
	select the target architecture

function	returnBatchOpt (BatchOptOut)
	return structure with Batch Options and possible configurations via the notify `syncBatch` event Parameters: BatchOptOut: a local structure with Batch Options generated during Continue callback. It may contain more fields than obj.BatchOpt structure

function	bioformatsCallback (event)
	update available filename extensions

function net =	selectNetwork (networkName)
	select a filename for a new network in the Train mode, or select a network to use for the Predict mode

function	selectDirerctories (event)

function	updateImageDirectoryPath (event)
	update directories with images for training, prediction and results

function	checkNetwork (fn)
	generate and check network using settings in the Train tab

function lgraph =	updateNetworkInputLayer (lgraph, inputPatchSize)
	update the input layer settings for lgraph paramters are taken from obj.InputLayerOpt

function lgraph =	updateActivationLayers (lgraph)
	update the activation layers depending on settings in obj.BatchOpt.T_ActivationLayer and obj.ActivationLayerOpt

function lgraph =	updateConvolutionLayers (lgraph)
	update the convolution layers by providing new set of weight initializers

function lgraph =	updateMaxPoolAndTransConvLayers (lgraph, poolSize)
	update maxPool and TransposedConvolution layers depending on network downsampling factor only for U-net and SegNet. This function is applied when the network downsampling factor is different from 2

function lgraph =	updateSegmentationLayer (lgraph, classNames)
	redefine the segmentation layer of lgraph based on obj.BatchOpt settings

function [ status , augNumber ] =	setAugFuncHandles (mode, augOptions)
	define list of 2D/3D augmentation functions

function [ dataOut , info ] =	classificationAugmentationPipeline (dataIn, info, inputPatchSize, outputPatchSize, mode)

function	setAugmentation3DSettings ()
	update settings for augmentation fo 3D images

function	setActivationLayerOptions ()
	update options for the activation layers

function	setSegmentationLayerOptions ()
	update options for the activation layers

function	setAugmentation2DSettings ()
	update settings for augmentation fo 2D images

function	setInputLayerSettings ()
	update init settings for the input layer of networks

function	setTrainingSettings ()
	update settings for training of networks

function	start (event)
	start calcualtions, depending on the selected tab preprocessing, training, or prediction is initialized

function imgOut =	channelWisePreProcess (imgIn)
	Normalize images As input has 4 channels (modalities), remove the mean and divide by the standard deviation of each modality independently.

function	processImages (preprocessFor)
	Preprocess images for training and prediction.

function	startPreprocessing ()

function TrainingOptions =	preprareTrainingOptions (valDS)
	prepare trainig options for the network training

function	updatePreprocessingMode ()
	callback for change of selection in the Preprocess for dropdown

function	saveConfig (configName)
	save Deep MIB configuration to a file

function res =	correctBatchOpt (res)
	correct loaded BatchOpt structure if it is not compatible with the current version of DeepMIB

function	loadConfig (configName)
	load config file with Deep MIB settings

function	startPredictionBlockedImage ()
	predict 2D/3D datasets using the blockedImage class requires R2021a or newer

function [ outputLabels , scoreImg ] =	processBlocksBlockedImage (vol, zValue, net, inputPatchSize, outputPatchSize, blockSize, padShift, dataDimension, patchwiseWorkflowSwitch, patchwisePatchesPredictSwitch, classNames, generateScoreFiles, executionEnvironment, fn)
	dataDimension: numeric switch that identify dataset dimension, can be 2, 2.5, 3

function	startPrediction2D ()
	predict datasets for 2D taken to a separate function for better performance

function	startPrediction3D ()
	predict datasets for 3D networks taken to a separate function to improve performance

function bls =	generateDynamicMaskingBlocks (vol, blockSize, noColors)
	generate blocks using dynamic masking parameters acquired in obj.DynamicMaskOpt

function	previewPredictions ()
	load images of prediction scores into MIB

function	previewModels (loadImagesSwitch)
	load images for predictions and the resulting modelsinto MIB

function	evaluateSegmentationPatches ()
	evaluate segmentation results for the patches in the patch-wise mode

function	evaluateSegmentation ()
	evaluate segmentation results by comparing predicted models with the ground truth models

function	selectGPUDevice ()
	select environment for computations

function	gpuInfo ()
	display information about the selected GPU

function	exportNetwork ()
	convert and export network to ONNX or TensorFlow formats

function	transferLearning ()
	perform fine-tuning of the loaded network to a different number of classes

function	importNetwork ()
	import an externally trained or designed network to be used with DeepMIB

function	updateDynamicMaskSettings ()
	update settings for calculation of dynamic masks during prediction using blockedimage mode the settings are stored in obj.DynamicMaskOpt

function	saveCheckpointNetworkCheck ()
	callback for press of Save checkpoint networks (obj.View.handles.T_SaveProgress)

function	previewDynamicMask ()
	preview results for the dynamic mode

function	exploreActivations ()
	explore activations within the trained network

function	countLabels ()
	count occurrences of labels in model files callback for press of the "Count labels" in the Options panel

function	balanceClasses ()
	balance classes before training see example from here: https://se.mathworks.com/help/vision/ref/balancepixellabels.html

function	customTrainingProgressWindow_Callback (event)
	callback for click on obj.View.handles.O_CustomTrainingProgressWindow checkbox

function	previewImagePatches_Callback (event)
	callback for value change of obj.View.handles.O_PreviewImagePatches

function	sendReportsCallback ()
	define parameters for sending progress report to the user's email address

function	helpButton_callback ()
	show Help sections

function	duplicateConfigAndNetwork ()
	copy the network file and its config to a new filename

Static Public Member Functions
static function	viewListner_Callback (obj, src, evnt)

static function data =	tif3DFileRead (filename)
	data = tif3DFileRead(filename) custom reading function to load tif files with stack of images used in evaluate segmentation function

static function [ outputLabeledImageBlock , scoreBlock ] =	segmentBlockedImage (block, net, dataDimension, patchwiseWorkflowSwitch, generateScoreFiles, executionEnvironment, padShift)
	test function for utilization of blockedImage for prediction The input block will be a batch of blocks from the a blockedImage.

Public Attributes
	mibModel
	handles to mibModel

	mibController
	handle to mib controller

	View
	handle to the view / mibDeepGUI

	listener
	a cell array with handles to listeners

	childControllers
	list of opened subcontrollers

	childControllersIds
	a cell array with names of initialized child controllers

	availableArchitectures
	containers.Map with available architectures keySet = {`2D Semantic`, `2.5D Semantic`, `3D Semantic`, `2D Patch-wise`, `2D Instance`}; valueSet{1} = {`DeepLab v3+`, `SegNet`, `U-net`, `U-net +Encoder`}; old: {`U-net`, `SegNet`, `DLv3 Resnet18`, `DLv3 Resnet50`, `DLv3 Xception`, `DLv3 Inception-ResNet-v2`} valueSet{2} = {`Z2C + DLv3`, `Z2C + DLv3`, `Z2C + U-net`, `Z2C + U-net +Encoder`}; % 3DC + DLv3 Resnet18' valueSet{3} = {`U-net`, `U-net Anisotropic`} valueSet{4} = {`Resnet18`, `Resnet50`, `Resnet101`, `Xception`} valueSet{5} = {`SOLOv2`} old: {`SOLOv2 Resnet18`, `SOLOv2 Resnet50`};

	availableEncoders
	containers.Map with available encoders keySet - is a mixture of workflow -space- architecture {`2D Semantic DeepLab v3+`, `2D Semantic U-net +Encoder`} encoders for each "workflow -space- architecture" combination, the last value shows the selected encoder encodersList{1} = {`Resnet18`, `Resnet50`, `Xception`, `InceptionResnetv2`, 1}; % for `2D Semantic DeepLab v3+` encodersList{2} = {`Classic`, `Resnet18`, `Resnet50`, 2}; % for `2D Semantic U-net +Encoder` encodersList{3} = {`Resnet18`, `Resnet50`, 1}; % for `2D Instance SOLOv2`'

	BatchOpt
	a structure compatible with batch operation name of each field should be displayed in a tooltip of GUI it is recommended that the Tags of widgets match the name of the fields in this structure .Parameter - [editbox], char/string .Checkbox - [checkbox], logical value true or false .Dropdown{1} - [dropdown], cell string for the dropdown .Dropdown{2} - [optional], an array with possible options .Radio - [radiobuttons], cell string `Radio1` or `Radio2`... .ParameterNumeric{1} - [numeric editbox], cell with a number .ParameterNumeric{2} - [optional], vector with limits [min, max] .ParameterNumeric{3} - [optional], string `on` - to round the value, `off` to do not round the value

	AugOpt2D
	a structure with augumentation options for 2D unets, default obtained from obj.mibModel.preferences.Deep.AugOpt2D, see getDefaultParameters.m .FillValue = 0; .RandXReflection = true; .RandYReflection = true; .RandRotation = [-10, 10]; .RandScale = [.95 1.05]; .RandXScale = [.95 1.05]; .RandYScale = [.95 1.05]; .RandXShear = [-5 5]; .RandYShear = [-5 5];

	Aug2DFuncNames
	cell array with names of 2D augmenter functions

	Aug2DFuncProbability
	probabilities of each 2D augmentation action to be triggered

	Aug3DFuncNames
	cell array with names of 2D augmenter functions

	Aug3DFuncProbability
	probabilities of each 3D augmentation action to be triggered

	gpuInfoFig
	a handle for GPU info window

	AugOpt3D
	.Fraction = .6; % augment 60% of patches .FillValue = 0; .RandXReflection = true; .RandYReflection = true; .RandZReflection = true; .Rotation90 = true; .ReflectedRotation90 = true;

	ActivationLayerOpt
	options for the activation layer

	DynamicMaskOpt
	options for calculation of dynamic masks for prediction using blocked image mode .Method = `Keep above threshold`; % `Keep above threshold` or `Keep below threshold` .ThresholdValue = 60; .InclusionThreshold = 0.1; % Inclusion threshold for mask blocks

	SegmentationLayerOpt
	options for the segmentation layer

	InputLayerOpt
	a structure with settings for the input layer .Normalization = `zerocenter`; .Mean = []; .StandardDeviation = []; .Min = []; .Max = [];

	modelMaterialColors
	colors of materials

	PatchPreviewOpt
	structure with preview patch options .noImages = 9, number of images in montage .imageSize = 160, patch size for preview .labelShow = true, display overlay labels with details .labelSize = 9, font size for the label .labelColor = `black`, color of the label .labelBgColor = `yellow`, color of the label background .labelBgOpacity = 0.6; % opacity of the background

	SendReports
	send email reports with progress of the training process .T_SendReports = false; .FROM_email = 'user@.nosp@m.gmai.nosp@m.l.com'; .SMTP_server = `smtp-relay.brevo.com`; .SMTP_port = `587`; .SMTP_auth = true; .SMTP_starttls = true; .SMTP_sername = 'user@.nosp@m.gmai.nosp@m.l.com'; .SMTP_password = '; .sendWhenFinished = false; .sendDuringRun = false;

	TrainingOpt
	a structure with training options, the default ones are obtained from obj.mibModel.preferences.Deep.TrainingOpt, see getDefaultParameters.m .solverName = `adam`; .MaxEpochs = 50; .Shuffle = `once`; .InitialLearnRate = 0.0005; .LearnRateSchedule = `piecewise`; .LearnRateDropPeriod = 10; .LearnRateDropFactor = 0.1; .L2Regularization = 0.0001; .Momentum = 0.9; .ValidationFrequency = 400; .Plots = `training-progress`;

	TrainingProgress
	a structure to be used for the training progress plot for the compiled version of MIB .maxNoIter = max number of iteractions for during training .iterPerEpoch - iterations per epoch .stopTraining - logical switch that forces to stop training and save all progress

	wb
	handle to waitbar

	colormap6
	colormap for 6 colors

	colormap20
	colormap for 20 colors

	colormap255
	colormap for 255 colors

	sessionSettings
	structure for the session settings .countLabelsDir - directory with labels to count, used in count labels function

	TrainEngine
	temp property to test trainnet function for training: can be `trainnet` or `trainNetwork`

EVENT	closeEvent
	> Description of events event firing when window is closed

Public Attributes inherited from handle
	addlistener
	Creates a listener for the specified event and assigns a callback function to execute when the event occurs.

	notify
	Broadcast a notice that a specific event is occurring on a specified handle object or array of handle objects.

	delete
	Handle object destructor method that is called when the object's lifecycle ends.

	disp
	Handle object disp method which is called by the display method. See the MATLAB disp function.

	display
	Handle object display method called when MATLAB software interprets an expression returning a handle object that is not terminated by a semicolon. See the MATLAB display function.

	findobj
	Finds objects matching the specified conditions from the input array of handle objects.

	findprop
	Returns a meta.property objects associated with the specified property name.

	fields
	Returns a cell array of string containing the names of public properties.

	fieldnames
	Returns a cell array of string containing the names of public properties. See the MATLAB fieldnames function.

	isvalid
	Returns a logical array in which elements are true if the corresponding elements in the input array are valid handles. This method is Sealed so you cannot override it in a handle subclass.

	eq
	Relational functions example. See details for more information.

	transpose
	Transposes the elements of the handle object array.

	permute
	Rearranges the dimensions of the handle object array. See the MATLAB permute function.

	reshape
	hanges the dimensions of the handle object array to the specified dimensions. See the MATLAB reshape function.

	sort
	ort the handle objects in any array in ascending or descending order.

Detailed Description

class is a template class for using with GUI developed using appdesigner of Matlab

obj.startController('mibDeepController'); // as GUI tool

or

// a code below was used for mibImageArithmeticController
BatchOpt.Parameter = 'test'mib;  // fill edit boxes as strings
BatchOpt.Checkbox = true;     // fill checkboxes with logicals: true/false
BatchOpt.Popup = {'value'};        // value for the popups as a cell
BatchOpt.Radio = {'Radio1'};          // selection of radio buttons, as cell with the handle of the target radio button
BatchOpt.showWaitbar = true;  // show or not the waitbar
obj.startController('mibDeepController', [], BatchOpt); // start mibDeepController in the batch mode

or

// trigger return of the possible Options using returnBatchOpt function
// using notify syncBatch event
obj.startController('mibDeepController', [], NaN);

Constructor & Destructor Documentation

◆ mibDeepController()

mibDeepController.mibDeepController	(		mibModel,
			varargin )

References handle.addlistener, DnD_uifigure(), gpuInfo(), mibModel, mibUpdateFontSize(), moveWindowOutside(), handle.notify, updateBatchOptCombineFields_Shared(), and updateGUIFromBatchOpt_Shared().

Referenced by evaluateSegmentation(), and processBlocksBlockedImage().

Here is the call graph for this function:

Here is the caller graph for this function:

Member Function Documentation

◆ activationLayerChangeCallback()

function mibDeepController.activationLayerChangeCallback ( )

callback for modification of the Activation Layer dropdown

◆ balanceClasses()

function mibDeepController.balanceClasses ( )

balance classes before training see example from here: https://se.mathworks.com/help/vision/ref/balancepixellabels.html

References mibInputMultiDlg(), and mibShowErrorDialog().

Here is the call graph for this function:

◆ bioformatsCallback()

function mibDeepController.bioformatsCallback ( event )

update available filename extensions

Parameters

event an event structure of appdesigner

Required fields of event:

◆ channelWisePreProcess()

function imgOut = mibDeepController.channelWisePreProcess ( imgIn )

Normalize images As input has 4 channels (modalities), remove the mean and divide by the standard deviation of each modality independently.

Parameters

imgIn input image, as matrix [heigth, width, color, depth]

Return values

imgOut resulting image, stretched between 0 and 1

References max, and min.

◆ checkNetwork()

function mibDeepController.checkNetwork ( fn )

generate and check network using settings in the Train tab

Parameters

fn	optional string with filename (*.mibDeep) to preview its configuration

References mibShowErrorDialog().

Here is the call graph for this function:

◆ classificationAugmentationPipeline()

function [ dataOut , info ] = mibDeepController.classificationAugmentationPipeline	(	dataIn,
		info,
		inputPatchSize,
		outputPatchSize,
		mode )

◆ closeWindow()

function mibDeepController.closeWindow ( )

update preferences structure

References handle.isvalid, and handle.notify.

◆ correctBatchOpt()

function res = mibDeepController.correctBatchOpt ( res )

correct loaded BatchOpt structure if it is not compatible with the current version of DeepMIB

Parameters

res	BatchOpt structure loaded from a file

Required fields of res:

Generated fields of res:

References mibDeepConvertOldAugmentationSettingsToNew().

Referenced by loadConfig().

Here is the call graph for this function:

Here is the caller graph for this function:

◆ countLabels()

function mibDeepController.countLabels ( )

count occurrences of labels in model files callback for press of the "Count labels" in the Options panel

References mibDeepStoreLoadCategorical(), mibDeepStoreLoadModel(), and mibInputMultiDlg().

Here is the call graph for this function:

◆ createNetwork()

function [ lgraph , outputPatchSize ] = mibDeepController.createNetwork ( previewSwitch )

generate network Parameters: previewSwitch: logical switch, when 1 - the generated network is only for preview, i.e. weights of classes won't be calculated

Return values

lgraph	network object
outputPatchSize	output patch size as [height, width, depth, color]

References mibShowErrorDialog().

Here is the call graph for this function:

◆ customTrainingProgressWindow_Callback()

function mibDeepController.customTrainingProgressWindow_Callback ( event )

callback for click on obj.View.handles.O_CustomTrainingProgressWindow checkbox

◆ dnd_files_callback()

function mibDeepController.dnd_files_callback	(		hWidget,
			dragIn )

drag and drop config name to obj.View.handles.NetworkPanel to load it

Parameters

hWidget	a handle to the object where the drag action landed
dragIn	a structure containing the dragged object .ctrlKey - 0/1 whether the control key was pressed .shiftKey - 0/1 whether the control key was pressed .names - cell array with filenames

Required fields of dragIn:

◆ duplicateConfigAndNetwork()

function mibDeepController.duplicateConfigAndNetwork ( )

copy the network file and its config to a new filename

References BatchOpt, mib_uigetfile(), and wb.

Here is the call graph for this function:

◆ evaluateSegmentation()

function mibDeepController.evaluateSegmentation ( )

evaluate segmentation results by comparing predicted models with the ground truth models

References handle.fieldnames, max, mibDeepController(), mibDeepStoreLoadImages(), mibDeepStoreLoadModel(), mibInputMultiDlg(), and mibShowErrorDialog().

Here is the call graph for this function:

◆ evaluateSegmentationPatches()

function mibDeepController.evaluateSegmentationPatches ( )

evaluate segmentation results for the patches in the patch-wise mode

References C().

Here is the call graph for this function:

◆ exploreActivations()

function mibDeepController.exploreActivations ( )

explore activations within the trained network

◆ exportNetwork()

function mibDeepController.exportNetwork ( )

convert and export network to ONNX or TensorFlow formats

References exportONNXNetwork(), mibInputMultiDlg(), mibShowErrorDialog(), and wb.

Here is the call graph for this function:

◆ generate3DDeepLabV3Network()

function net = mibDeepController.generate3DDeepLabV3Network	(	imageSize,
		numClasses,
		downsamplingFactor,
		targetNetwork )

generate a hybrid 2.5D DeepLabv3+ convolutional neural network for semantic image segmentation. The training data should be a small substack of 3,5,7 etc slices, where only the middle slice is segmented. As the update, a 2 blocks of 3D convolutions were added before the standard DLv3

Parameters

imageSize	vector [height, width, colors] defining input patch size, should be larger than [224 224] for resnet18, colors should be 3
numClasses	number of output classes (including exterior) for the output results
targetNetwork	string defining the base architecture for the initialization `resnet18` - resnet18 network `resnet50` - resnet50 network `xception` - xception network `inceptionresnetv2` - resnet50 network

References mibInputMultiDlg().

Here is the call graph for this function:

◆ generateDeepLabV3Network()

function net = mibDeepController.generateDeepLabV3Network	(	imageSize,
		numClasses,
		targetNetwork )

generate DeepLab v3+ convolutional neural network for semantic image segmentation of 2D RGB images

Parameters

imageSize	vector [height, width, colors] defining input patch size, should be larger than [224 224] for resnet18, colors should be 3
numClasses	number of output classes (including exterior) for the output results
targetNetwork	string defining the base architecture for the initialization `resnet18` - resnet18 network `resnet50` - resnet50 network `xception` - xception network (requires matlab) `inceptionresnetv2` - inceptionresnetv2 network (required matlab)

References max, and mibInputMultiDlg().

Here is the call graph for this function:

◆ generateDynamicMaskingBlocks()

function bls = mibDeepController.generateDynamicMaskingBlocks	(	vol,
		blockSize,
		noColors )

generate blocks using dynamic masking parameters acquired in obj.DynamicMaskOpt

Parameters

vol	blocked image to process
blockSize	block size
noColors	number of color channels in the blocked image

Return values

bls	calculated blockLocationSet .ImageNumber .BlockOrigin .BlockSize .Levels

References max.

◆ generateUnet2DwithEncoder()

function [ net , outputSize ] = mibDeepController.generateUnet2DwithEncoder	(		imageSize,
			encoderNetwork )

generate Unet convolutional neural network for semantic image segmentation of 2D RGB images using a specified encoder

Parameters

imageSize	vector [height, width, colors] defining input patch size, should be larger than [224 224] for Resnet18, colors should be 3
encoderNetwork	string defining the base architecture for the initialization `Classic` - classic unet architecture `Resnet18` - Resnet18 network `Resnet50` - Resnet50 network

Return values

net	Unet dlnetwork, with softmax (Name: `FinalNetworkSoftmax-Layer`) as the final layer
outputSize	output size of the network returned as [height, width, number of classes] Updates

◆ gpuInfo()

function mibDeepController.gpuInfo ( )

display information about the selected GPU

References D(), and handle.fieldnames.

Referenced by mibDeepController().

Here is the call graph for this function:

Here is the caller graph for this function:

◆ helpButton_callback()

function mibDeepController.helpButton_callback ( )

show Help sections

◆ importNetwork()

function mibDeepController.importNetwork ( )

import an externally trained or designed network to be used with DeepMIB

Example: %

generate a network: net = deeplabv3plusLayers([512 512 3], 5, resnet18); % save network to a file save(myNewNetwork.mat, net, -mat); % use Import opetation to load and adapt the network for use with DeepMIB

References ActivationLayerOpt, BatchOpt, DynamicMaskOpt, handle.fieldnames, InputLayerOpt, mib_uigetfile(), mibInputMultiDlg(), SegmentationLayerOpt, and wb.

Here is the call graph for this function:

◆ loadConfig()

function mibDeepController.loadConfig ( configName )

load config file with Deep MIB settings

Parameters

configName full filename for the config file to load

References correctBatchOpt(), mib_uigetfile(), mibConcatenateStructures(), and updateBatchOptCombineFields_Shared().

Here is the call graph for this function:

◆ mibDeepAugmentAndCrop2dPatchMultiGPU()

function [ patchOut , info , augList , augPars ] = mibDeepController.mibDeepAugmentAndCrop2dPatchMultiGPU	(	info,
		inputPatchSize,
		outputPatchSize,
		mode,
		options )

◆ mibDeepAugmentAndCrop3dPatchMultiGPU()

function [ patchOut , info , augList , augPars ] = mibDeepController.mibDeepAugmentAndCrop3dPatchMultiGPU	(	info,
		inputPatchSize,
		outputPatchSize,
		mode,
		options )

◆ preprareTrainingOptions()

function TrainingOptions = mibDeepController.preprareTrainingOptions ( valDS )

prepare trainig options for the network training

Parameters

valDS datastore with images for validation

References mibShowErrorDialog().

Here is the call graph for this function:

◆ preprareTrainingOptionsInstances()

function TrainingOptions = mibDeepController.preprareTrainingOptionsInstances ( valDS )

prepare trainig options for training of the instance segmentation network

Parameters

valDS datastore with images for validation

References mibShowErrorDialog().

Here is the call graph for this function:

◆ previewDynamicMask()

function mibDeepController.previewDynamicMask ( )

preview results for the dynamic mode

References wb.

◆ previewImagePatches_Callback()

function mibDeepController.previewImagePatches_Callback ( event )

callback for value change of obj.View.handles.O_PreviewImagePatches

◆ previewModels()

function mibDeepController.previewModels ( loadImagesSwitch )

load images for predictions and the resulting modelsinto MIB

Parameters

loadImagesSwitch [logical], load or not (assuming that images have already been preloaded) images. When true, both images and models are loaded, when false - only models are loaded

◆ previewPredictions()

function mibDeepController.previewPredictions ( )

load images of prediction scores into MIB

◆ processBlocksBlockedImage()

function [ outputLabels , scoreImg ] = mibDeepController.processBlocksBlockedImage	(	vol,
		zValue,
		net,
		inputPatchSize,
		outputPatchSize,
		blockSize,
		padShift,
		dataDimension,
		patchwiseWorkflowSwitch,
		patchwisePatchesPredictSwitch,
		classNames,
		generateScoreFiles,
		executionEnvironment,
		fn )

dataDimension: numeric switch that identify dataset dimension, can be 2, 2.5, 3

Required fields of vol:

References max, mibDeepController(), and handle.permute.

Here is the call graph for this function:

◆ processImages()

function mibDeepController.processImages ( preprocessFor )

Preprocess images for training and prediction.

Parameters

preprocessFor a string with target, training, prediction

References amiraMesh2bitmap(), mibDeepStoreLoadImages(), mibDeepStoreLoadModel(), mibLoadImages(), mibShowErrorDialog(), handle.permute, and saveImageParFor().

Here is the call graph for this function:

◆ returnBatchOpt()

function mibDeepController.returnBatchOpt ( BatchOptOut )

return structure with Batch Options and possible configurations via the notify syncBatch event Parameters: BatchOptOut: a local structure with Batch Options generated during Continue callback. It may contain more fields than obj.BatchOpt structure

References handle.notify.

◆ saveCheckpointNetworkCheck()

function mibDeepController.saveCheckpointNetworkCheck ( )

callback for press of Save checkpoint networks (obj.View.handles.T_SaveProgress)

References mibInputMultiDlg().

Here is the call graph for this function:

◆ saveConfig()

function mibDeepController.saveConfig ( configName )

save Deep MIB configuration to a file

Parameters

configName [optional] string, full filename to the config file

References ActivationLayerOpt, BatchOpt, convertAbsoluteToRelativePath(), DynamicMaskOpt, InputLayerOpt, and SegmentationLayerOpt.

Here is the call graph for this function:

◆ segmentBlockedImage()

static function [ outputLabeledImageBlock , scoreBlock ] = mibDeepController.segmentBlockedImage	(	block,
		net,
		dataDimension,
		patchwiseWorkflowSwitch,
		generateScoreFiles,
		executionEnvironment,
		padShift )

static

test function for utilization of blockedImage for prediction The input block will be a batch of blocks from the a blockedImage.

  .BlockSub: [1 1 1]
  .Start: [1 1 1]
  .End: [224 224 3]
  .Level: 1
  .ImageNumber: 1
  .BorderSize: [0 0 0]
  .BlockSize: [224 224 3]
  .BatchSize: 1
  .Data: [224×224×3 uint8]

net: a trained DAGNetwork dataDimension: numeric switch that identify dataset dimension, can be 2, 2.5, 3 patchwiseWorkflowSwitch: logical switch indicating the patch-wise mode, when true->use patch mode, when false->use semantic segmentation generateScoreFiles: variable to generate score files with probabilities of classes 0-> do not generate 1-> Use AM format 2-> Use Matlab non-compressed format 3-> Use Matlab compressed format 4-> Use Matlab non-compressed format (range 0-1) executionEnvironment: string with the environment to execute prediction padShift: numeric, (y,x,z or y,x) value for the padding, used during the overlap mode to crop the output patch for export

Parameters

block a structure with a block that is provided by blockedImage/apply. The first and second iterations have batch size==1, while the following have the batch size equal to the selected. Below fields of the structure,

Required fields of block:

References handle.permute, and handle.reshape.

◆ selectArchitecture()

function mibDeepController.selectArchitecture ( event )

select the target architecture

◆ selectDirerctories()

function mibDeepController.selectDirerctories ( event )

◆ selectGPUDevice()

function mibDeepController.selectGPUDevice ( )

select environment for computations

◆ selectNetwork()

function net = mibDeepController.selectNetwork ( networkName )

select a filename for a new network in the Train mode, or select a network to use for the Predict mode

Parameters

networkName optional parameter with the network full filename

Return values

net	trained network

References handle.fieldnames, mib_uigetfile(), mibConcatenateStructures(), and updateBatchOptCombineFields_Shared().

Here is the call graph for this function:

◆ selectWorkflow()

function mibDeepController.selectWorkflow ( event )

select deep learning workflow to perform

◆ sendReportsCallback()

function mibDeepController.sendReportsCallback ( )

define parameters for sending progress report to the user's email address

References mibInputMultiDlg().

Here is the call graph for this function:

◆ setActivationLayerOptions()

function mibDeepController.setActivationLayerOptions ( )

update options for the activation layers

References mibInputMultiDlg().

Here is the call graph for this function:

◆ setAugFuncHandles()

function [ status , augNumber ] = mibDeepController.setAugFuncHandles	(		mode,
			augOptions )

define list of 2D/3D augmentation functions

Parameters

mode	string defining `2D` or `3D` augmentations
augOptions	a custom temporary structure with augmentation options to be used instead of obj.AugOpt2D and obj.AugOpt3D. It is used by mibDeepAugmentSettingsController to preview selected augmentations

Return values

status	a logical success switch (1-success, 0- fail)
augNumber	number of selected augmentations

References handle.fieldnames.

◆ setAugmentation2DSettings()

function mibDeepController.setAugmentation2DSettings ( )

update settings for augmentation fo 2D images

References mibDeepConvertOldAugmentationSettingsToNew().

Here is the call graph for this function:

◆ setAugmentation3DSettings()

function mibDeepController.setAugmentation3DSettings ( )

update settings for augmentation fo 3D images

References mibDeepConvertOldAugmentationSettingsToNew().

Here is the call graph for this function:

◆ setInputLayerSettings()

function mibDeepController.setInputLayerSettings ( )

update init settings for the input layer of networks

References mibInputMultiDlg(), and handle.reshape.

Here is the call graph for this function:

◆ setSegmentationLayer()

function mibDeepController.setSegmentationLayer ( )

callback for modification of the Segmentation Layer dropdown

◆ setSegmentationLayerOptions()

function mibDeepController.setSegmentationLayerOptions ( )

update options for the activation layers

References mibInputMultiDlg().

Here is the call graph for this function:

◆ setTrainingSettings()

function mibDeepController.setTrainingSettings ( )

update settings for training of networks

References mibInputMultiDlg().

Here is the call graph for this function:

◆ singleModelTrainingFileValueChanged()

function mibDeepController.singleModelTrainingFileValueChanged ( event )

callback for press of SingleModelTrainingFile

◆ start()

function mibDeepController.start ( event )

start calcualtions, depending on the selected tab preprocessing, training, or prediction is initialized

Required fields of event:

References mibInputMultiDlg(), and mibShowErrorDialog().

Here is the call graph for this function:

◆ startPrediction2D()

function mibDeepController.startPrediction2D ( )

predict datasets for 2D taken to a separate function for better performance

References bitmap2amiraMesh(), max, mibDeepStoreLoadImages(), mibShowErrorDialog(), modelMaterialColors, handle.notify, and saveImageParFor().

Here is the call graph for this function:

◆ startPrediction3D()

function mibDeepController.startPrediction3D ( )

predict datasets for 3D networks taken to a separate function to improve performance

References bitmap2amiraMesh(), max, mibDeepStoreLoadImages(), mibShowErrorDialog(), min, modelMaterialColors, handle.notify, handle.permute, and saveImageParFor().

Here is the call graph for this function:

◆ startPredictionBlockedImage()

function mibDeepController.startPredictionBlockedImage ( )

predict 2D/3D datasets using the blockedImage class requires R2021a or newer

References bitmap2amiraMesh(), max, mibDeepStoreLoadImages(), mibDoImageFiltering(), mibResize3d(), mibShowErrorDialog(), min, modelMaterialColors, handle.notify, and saveImageParFor().

Here is the call graph for this function:

◆ startPreprocessing()

function mibDeepController.startPreprocessing ( )

References mibShowErrorDialog(), and wb.

Here is the call graph for this function:

◆ tif3DFileRead()

static function data = mibDeepController.tif3DFileRead ( filename )

static

data = tif3DFileRead(filename) custom reading function to load tif files with stack of images used in evaluate segmentation function

◆ toggleAugmentations()

function mibDeepController.toggleAugmentations ( )

callback for press of the T_augmentation checkbox

◆ transferLearning()

function mibDeepController.transferLearning ( )

perform fine-tuning of the loaded network to a different number of classes

References mibInputMultiDlg().

Here is the call graph for this function:

◆ updateActivationLayers()

function lgraph = mibDeepController.updateActivationLayers ( lgraph )

update the activation layers depending on settings in obj.BatchOpt.T_ActivationLayer and obj.ActivationLayerOpt

Required fields of lgraph:

◆ updateBatchOptFromGUI()

function mibDeepController.updateBatchOptFromGUI ( event )

References updateBatchOptFromGUI_Shared().

Here is the call graph for this function:

◆ updateConvolutionLayers()

function lgraph = mibDeepController.updateConvolutionLayers ( lgraph )

update the convolution layers by providing new set of weight initializers

Required fields of lgraph:

◆ updateDynamicMaskSettings()

function mibDeepController.updateDynamicMaskSettings ( )

update settings for calculation of dynamic masks during prediction using blockedimage mode the settings are stored in obj.DynamicMaskOpt

References mibInputMultiDlg().

Here is the call graph for this function:

◆ updateImageDirectoryPath()

function mibDeepController.updateImageDirectoryPath ( event )

update directories with images for training, prediction and results

Required fields of event:

◆ updateMaxPoolAndTransConvLayers()

function lgraph = mibDeepController.updateMaxPoolAndTransConvLayers	(		lgraph,
			poolSize )

update maxPool and TransposedConvolution layers depending on network downsampling factor only for U-net and SegNet. This function is applied when the network downsampling factor is different from 2

Required fields of lgraph:

◆ updateNetworkInputLayer()

function lgraph = mibDeepController.updateNetworkInputLayer	(		lgraph,
			inputPatchSize )

update the input layer settings for lgraph paramters are taken from obj.InputLayerOpt

Required fields of lgraph:

References handle.reshape.

◆ updatePreprocessingMode()

function mibDeepController.updatePreprocessingMode ( )

callback for change of selection in the Preprocess for dropdown

◆ updateSegmentationLayer()

function lgraph = mibDeepController.updateSegmentationLayer	(		lgraph,
			classNames )

redefine the segmentation layer of lgraph based on obj.BatchOpt settings

Parameters

classNames cell array with class names, when not provided is auto switch is used

Required fields of lgraph:

◆ updateWidgets()

function mibDeepController.updateWidgets ( )

update widgets of this window

References updateGUIFromBatchOpt_Shared().

Here is the call graph for this function:

◆ viewListner_Callback()

static function mibDeepController.viewListner_Callback	(	obj,
		src,
		evnt )

static

Member Data Documentation

◆ ActivationLayerOpt

mibDeepController.ActivationLayerOpt

options for the activation layer

Referenced by importNetwork(), and saveConfig().

◆ Aug2DFuncNames

mibDeepController.Aug2DFuncNames

cell array with names of 2D augmenter functions

◆ Aug2DFuncProbability

mibDeepController.Aug2DFuncProbability

probabilities of each 2D augmentation action to be triggered

◆ Aug3DFuncNames

mibDeepController.Aug3DFuncNames

cell array with names of 2D augmenter functions

◆ Aug3DFuncProbability

mibDeepController.Aug3DFuncProbability

probabilities of each 3D augmentation action to be triggered

◆ AugOpt2D

mibDeepController.AugOpt2D

a structure with augumentation options for 2D unets, default obtained from obj.mibModel.preferences.Deep.AugOpt2D, see getDefaultParameters.m .FillValue = 0; .RandXReflection = true; .RandYReflection = true; .RandRotation = [-10, 10]; .RandScale = [.95 1.05]; .RandXScale = [.95 1.05]; .RandYScale = [.95 1.05]; .RandXShear = [-5 5]; .RandYShear = [-5 5];

◆ AugOpt3D

mibDeepController.AugOpt3D

.Fraction = .6; % augment 60% of patches .FillValue = 0; .RandXReflection = true; .RandYReflection = true; .RandZReflection = true; .Rotation90 = true; .ReflectedRotation90 = true;

◆ availableArchitectures

mibDeepController.availableArchitectures

containers.Map with available architectures keySet = {2D Semantic, 2.5D Semantic, 3D Semantic, 2D Patch-wise, 2D Instance}; valueSet{1} = {DeepLab v3+, SegNet, U-net, U-net +Encoder}; old: {U-net, SegNet, DLv3 Resnet18, DLv3 Resnet50, DLv3 Xception, DLv3 Inception-ResNet-v2} valueSet{2} = {Z2C + DLv3, Z2C + DLv3, Z2C + U-net, Z2C + U-net +Encoder}; % 3DC + DLv3 Resnet18' valueSet{3} = {U-net, U-net Anisotropic} valueSet{4} = {Resnet18, Resnet50, Resnet101, Xception} valueSet{5} = {SOLOv2} old: {SOLOv2 Resnet18, SOLOv2 Resnet50};

◆ availableEncoders

mibDeepController.availableEncoders

containers.Map with available encoders keySet - is a mixture of workflow -space- architecture {2D Semantic DeepLab v3+, 2D Semantic U-net +Encoder} encoders for each "workflow -space- architecture" combination, the last value shows the selected encoder encodersList{1} = {Resnet18, Resnet50, Xception, InceptionResnetv2, 1}; % for 2D Semantic DeepLab v3+ encodersList{2} = {Classic, Resnet18, Resnet50, 2}; % for 2D Semantic U-net +Encoder encodersList{3} = {Resnet18, Resnet50, 1}; % for 2D Instance SOLOv2'

◆ BatchOpt

mibDeepController.BatchOpt

a structure compatible with batch operation name of each field should be displayed in a tooltip of GUI it is recommended that the Tags of widgets match the name of the fields in this structure .Parameter - [editbox], char/string .Checkbox - [checkbox], logical value true or false .Dropdown{1} - [dropdown], cell string for the dropdown .Dropdown{2} - [optional], an array with possible options .Radio - [radiobuttons], cell string Radio1 or Radio2... .ParameterNumeric{1} - [numeric editbox], cell with a number .ParameterNumeric{2} - [optional], vector with limits [min, max] .ParameterNumeric{3} - [optional], string on - to round the value, off to do not round the value

Referenced by duplicateConfigAndNetwork(), importNetwork(), and saveConfig().

◆ childControllers

mibDeepController.childControllers

list of opened subcontrollers

◆ childControllersIds

mibDeepController.childControllersIds

a cell array with names of initialized child controllers

◆ closeEvent

EVENT mibDeepController.closeEvent

> Description of events event firing when window is closed

Events: closeEvent

◆ colormap20

mibDeepController.colormap20

colormap for 20 colors

◆ colormap255

mibDeepController.colormap255

colormap for 255 colors

◆ colormap6

mibDeepController.colormap6

colormap for 6 colors

◆ DynamicMaskOpt

mibDeepController.DynamicMaskOpt

options for calculation of dynamic masks for prediction using blocked image mode .Method = Keep above threshold; % Keep above threshold or Keep below threshold .ThresholdValue = 60; .InclusionThreshold = 0.1; % Inclusion threshold for mask blocks

Referenced by importNetwork(), and saveConfig().

◆ gpuInfoFig

mibDeepController.gpuInfoFig

a handle for GPU info window

◆ InputLayerOpt

mibDeepController.InputLayerOpt

a structure with settings for the input layer .Normalization = zerocenter; .Mean = []; .StandardDeviation = []; .Min = []; .Max = [];

Referenced by importNetwork(), and saveConfig().

◆ listener

mibDeepController.listener

a cell array with handles to listeners

◆ mibController

mibDeepController.mibController

handle to mib controller

◆ mibModel

mibDeepController.mibModel

handles to mibModel

Referenced by mibDeepController().

◆ modelMaterialColors

mibDeepController.modelMaterialColors

colors of materials

Referenced by startPrediction2D(), startPrediction3D(), and startPredictionBlockedImage().

◆ PatchPreviewOpt

mibDeepController.PatchPreviewOpt

structure with preview patch options .noImages = 9, number of images in montage .imageSize = 160, patch size for preview .labelShow = true, display overlay labels with details .labelSize = 9, font size for the label .labelColor = black, color of the label .labelBgColor = yellow, color of the label background .labelBgOpacity = 0.6; % opacity of the background

◆ SegmentationLayerOpt

mibDeepController.SegmentationLayerOpt

options for the segmentation layer

Referenced by importNetwork(), and saveConfig().

◆ SendReports

mibDeepController.SendReports

send email reports with progress of the training process .T_SendReports = false; .FROM_email = 'user@.nosp@m.gmai.nosp@m.l.com'; .SMTP_server = smtp-relay.brevo.com; .SMTP_port = 587; .SMTP_auth = true; .SMTP_starttls = true; .SMTP_sername = 'user@.nosp@m.gmai.nosp@m.l.com'; .SMTP_password = '; .sendWhenFinished = false; .sendDuringRun = false;

◆ sessionSettings

mibDeepController.sessionSettings

structure for the session settings .countLabelsDir - directory with labels to count, used in count labels function

◆ TrainEngine

mibDeepController.TrainEngine

temp property to test trainnet function for training: can be trainnet or trainNetwork

◆ TrainingOpt

mibDeepController.TrainingOpt

a structure with training options, the default ones are obtained from obj.mibModel.preferences.Deep.TrainingOpt, see getDefaultParameters.m .solverName = adam; .MaxEpochs = 50; .Shuffle = once; .InitialLearnRate = 0.0005; .LearnRateSchedule = piecewise; .LearnRateDropPeriod = 10; .LearnRateDropFactor = 0.1; .L2Regularization = 0.0001; .Momentum = 0.9; .ValidationFrequency = 400; .Plots = training-progress;

◆ TrainingProgress

mibDeepController.TrainingProgress

a structure to be used for the training progress plot for the compiled version of MIB .maxNoIter = max number of iteractions for during training .iterPerEpoch - iterations per epoch .stopTraining - logical switch that forces to stop training and save all progress

◆ View

mibDeepController.View

handle to the view / mibDeepGUI

◆ wb

mibDeepController.wb

handle to waitbar

Referenced by duplicateConfigAndNetwork(), exportNetwork(), importNetwork(), previewDynamicMask(), and startPreprocessing().

The documentation for this class was generated from the following files:

Classes/@mibDeepController/mibDeepController.m
Classes/@mibDeepController/createNetwork.m
Classes/@mibDeepController/generate3DDeepLabV3Network.m
Classes/@mibDeepController/generateDeepLabV3Network.m
Classes/@mibDeepController/generateUnet2DwithEncoder.m
Classes/@mibDeepController/preprareTrainingOptionsInstances.m

Public Member Functions

Static Public Member Functions

Public Attributes

Detailed Description

Constructor & Destructor Documentation

◆ mibDeepController()

Member Function Documentation

◆ activationLayerChangeCallback()

◆ balanceClasses()

◆ bioformatsCallback()

◆ channelWisePreProcess()

◆ checkNetwork()

◆ classificationAugmentationPipeline()

◆ closeWindow()

◆ correctBatchOpt()

◆ countLabels()

◆ createNetwork()

◆ customTrainingProgressWindow_Callback()

◆ dnd_files_callback()

◆ duplicateConfigAndNetwork()

◆ evaluateSegmentation()

◆ evaluateSegmentationPatches()

◆ exploreActivations()

◆ exportNetwork()

◆ generate3DDeepLabV3Network()

◆ generateDeepLabV3Network()

◆ generateDynamicMaskingBlocks()

◆ generateUnet2DwithEncoder()

◆ gpuInfo()

◆ helpButton_callback()

◆ importNetwork()

◆ loadConfig()

◆ mibDeepAugmentAndCrop2dPatchMultiGPU()

◆ mibDeepAugmentAndCrop3dPatchMultiGPU()

◆ preprareTrainingOptions()

◆ preprareTrainingOptionsInstances()

◆ previewDynamicMask()

◆ previewImagePatches_Callback()

◆ previewModels()

◆ previewPredictions()

◆ processBlocksBlockedImage()

◆ processImages()

◆ returnBatchOpt()

◆ saveCheckpointNetworkCheck()

◆ saveConfig()

◆ segmentBlockedImage()

◆ selectArchitecture()

◆ selectDirerctories()

◆ selectGPUDevice()

◆ selectNetwork()

◆ selectWorkflow()

◆ sendReportsCallback()

◆ setActivationLayerOptions()

◆ setAugFuncHandles()

◆ setAugmentation2DSettings()

◆ setAugmentation3DSettings()

◆ setInputLayerSettings()

◆ setSegmentationLayer()

◆ setSegmentationLayerOptions()

◆ setTrainingSettings()

◆ singleModelTrainingFileValueChanged()

◆ start()

◆ startPrediction2D()

◆ startPrediction3D()

◆ startPredictionBlockedImage()

◆ startPreprocessing()

◆ tif3DFileRead()

◆ toggleAugmentations()

◆ transferLearning()

◆ updateActivationLayers()

◆ updateBatchOptFromGUI()

◆ updateConvolutionLayers()

◆ updateDynamicMaskSettings()

◆ updateImageDirectoryPath()

◆ updateMaxPoolAndTransConvLayers()

◆ updateNetworkInputLayer()

◆ updatePreprocessingMode()

◆ updateSegmentationLayer()

◆ updateWidgets()

◆ viewListner_Callback()