de.unihalle.informatik.MiToBo.apps.cytoskeleton

Class CytoskeletonAnalyzer2D

java.lang.Object

de.unihalle.informatik.Alida.operator.ALDOperator

de.unihalle.informatik.MiToBo.core.operator.MTBOperator

de.unihalle.informatik.MiToBo.apps.cytoskeleton.CytoskeletonAnalyzer2D

All Implemented Interfaces:

de.unihalle.informatik.Alida.datatypes.ALDConfigurationValidator, de.unihalle.informatik.Alida.operator.events.ALDOperatorExecutionProgressEventListener, EventListener

@ALDAOperator(genericExecutionMode=ALL,
              level=APPLICATION,
              allowBatchMode=false,
              shortDescription="Extracts quantative global features of cytoskeletons.")
public class CytoskeletonAnalyzer2D
extends MTBOperator

Operator for extracting quantative global features of the cytoskeleton.

Author:

moeller

Nested Class Summary

Nested classes/interfaces inherited from class de.unihalle.informatik.Alida.operator.ALDOperator

de.unihalle.informatik.Alida.operator.ALDOperator.HidingMode, de.unihalle.informatik.Alida.operator.ALDOperator.HistoryConstructionMode

Field Summary

Fields

Modifier and Type	Field and Description
protected Vector<org.jfree.chart.JFreeChart>	boxWhiskerCharts Box-whisker plot of the group-wise cluster distributions.
private TreeSet<String>	cellGroupNames List of group names, filled in clusterFeatures().
private Vector<HashMap<String,HashMap<String,Double>>>	cellGroups Group-wise cell distribution data, for each group the vector contains a hash map with the following structure: - the keys of the map are given by the cell IDs of the different cells (without basename which is equal to the group name) - the values are represented again as a hash map where the keys are given by the cluster IDs and the values by the relative frequency of the corresponding cluster in the cell Example: --------- cellGroups.get(0) = < 001-01, < [c1, vc1], ..., [c6, vc6] > > ,
private LinkedList<String>	cellwiseDistroKeys Sorted list of keys in cellwiseDistros once it is filled.
private HashMap<String,double[]>	cellwiseDistros Map of cluster distributions per cell, filled in clusterFeatures().
protected int	clusterNum Number of clusters to be used in feature clustering.
private int	cytoSkelChannel Channel of input image containing stained cytoskeleton.
private double[][]	distroData Cluster distribution data, rows refer to clusters and columns to cells.
private String[]	distroDataGroups Group names linked to entries in distroData.
protected boolean	doFeatureCalculation Flag for calculating features.
protected boolean	doPCA Perform PCA in second stage?
protected CytoskeletonFeatureExtractor	featureExtractor Feature extractor to apply.
private static String	fileSep File separator character, operating system dependent.
protected de.unihalle.informatik.Alida.datatypes.ALDDirectoryString	imageDir Input image directory.
private int	imageHeight Height of the images taking first image as reference.
private int	imageWidth Width of the images taking first image as reference.
protected de.unihalle.informatik.Alida.datatypes.ALDDirectoryString	maskDir (Optional) directory with (cell) boundaries.
protected CytoskeletonFeatureExtractor.CellMaskFormat	maskFormat Format of provided cell boundaries.
private static String	operatorID Identifier string for this operator class.
protected Vector<org.jfree.chart.JFreeChart>	stackedBarCharts Resulting stacked bar plot of cluster distributions.
private double[][]	subspaceData Dimension-reduced cluster distribution data.
protected int	tileShiftX Tile shift in x-direction.
protected int	tileShiftY Tile size in y-direction.
protected int	tileSizeX Tile size in x-direction.
protected int	tileSizeY Tile size in y-direction.

Fields inherited from class de.unihalle.informatik.Alida.operator.ALDOperator

completeDAG, name, operatorExecutionEventlistenerList, portHashAccess, verbose, versionProvider

Constructor Summary

Constructors

Constructor and Description
CytoskeletonAnalyzer2D() Default constructor.

Method Summary

Modifier and Type	Method and Description
private void	calcFeatureFlagChanged() Callback routine to change parameters on change of flag for enable/disable feature calculation.
private void	calculateDistanceMatrices() Calculates pairwise distance matrices and similarity network data.
private void	clusterFeatures() Performs the feature clustering via Weka's kMeans algorithm.
private void	doPCA() Performs a Karhunen-Loeve transformation on the cluster distribution data.
String	getDocumentation()
protected void	operate()

Methods inherited from class de.unihalle.informatik.MiToBo.core.operator.MTBOperator

readResolve

Methods inherited from class de.unihalle.informatik.Alida.operator.ALDOperator

addOperatorExecutionProgressEventListener, addParameter, addParameter, addParameterUnconditioned, fieldContained, fireOperatorExecutionProgressEvent, getALDPortHashAccessKey, getConstructionMode, getHidingMode, getInactiveParameterNames, getInInoutNames, getInInoutNames, getInNames, getInOutNames, getMissingRequiredInputs, getName, getNumParameters, getOutInoutNames, getOutNames, getParameter, getParameterDescriptor, getParameterDescriptorUnconditioned, getParameterNames, getParameterUnconditioned, getSupplementalNames, getVerbose, getVersion, handleOperatorExecutionProgressEvent, hasInOutParameters, hasParameter, isAnnotatedParameter, isConfigured, print, print, print, printInterface, printInterface, readHistory, reinitializeParameterDescriptors, removeOperatorExecutionProgressEventListener, removeParameter, runOp, runOp, runOp, setConstructionMode, setConstructionMode, setConstructionMode, setHidingMode, setName, setParameter, setParameterUnconditioned, setVerbose, toStringVerbose, unconfiguredItems, validate, validateCustom, validateGeneric, writeHistory, writeHistory, writeHistory

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

operatorID

private static final String operatorID

Identifier string for this operator class.

See Also:

Constant Field Values

imageDir

@Parameter(label="Image file folder",
           required=true,
           dataIOOrder=-10,
           direction=IN,
           description="Input image directory.",
           mode=STANDARD)
protected de.unihalle.informatik.Alida.datatypes.ALDDirectoryString imageDir

Input image directory.

All files in the directory are considered. If a file cannot be opened (e.g. because it is not an image) it is skipped.

maskFormat

@Parameter(label="Boundary file format",
           required=true,
           dataIOOrder=-8,
           direction=IN,
           description="Format of cell boundary files.",
           mode=STANDARD)
protected CytoskeletonFeatureExtractor.CellMaskFormat maskFormat

Format of provided cell boundaries.

cytoSkelChannel

@Parameter(label="Cytoskeleton channel",
           required=true,
           direction=IN,
           dataIOOrder=-7,
           description="Channel with stained cytoskeleton, e.g., 1, 2 and so on.")
private int cytoSkelChannel

Channel of input image containing stained cytoskeleton.

doFeatureCalculation

@Parameter(label="Calculate features?",
           required=true,
           dataIOOrder=0,
           direction=IN,
           description="Flag to enable/disable feature calculation.",
           mode=STANDARD,
           callback="calcFeatureFlagChanged",
           paramModificationMode=MODIFIES_INTERFACE)
protected boolean doFeatureCalculation

Flag for calculating features.

If false, features are assumed to be available already.

featureExtractor

@Parameter(label="Feature Extractor",
           required=true,
           dataIOOrder=1,
           direction=IN,
           description="Select type of features to apply.",
           mode=STANDARD)
protected CytoskeletonFeatureExtractor featureExtractor

Feature extractor to apply.

tileSizeX

@Parameter(label="Tile size x",
           required=true,
           dataIOOrder=3,
           direction=IN,
           mode=STANDARD,
           description="Tile size in x-direction.")
protected int tileSizeX

Tile size in x-direction.

tileSizeY

@Parameter(label="Tile size y",
           required=true,
           dataIOOrder=4,
           direction=IN,
           mode=STANDARD,
           description="Tile size in y-direction.")
protected int tileSizeY

Tile size in y-direction.

tileShiftX

@Parameter(label="Tile shift x",
           required=true,
           dataIOOrder=5,
           direction=IN,
           mode=STANDARD,
           description="Tile shift in x-direction.")
protected int tileShiftX

Tile shift in x-direction.

tileShiftY

@Parameter(label="Tile shift y",
           required=true,
           dataIOOrder=6,
           direction=IN,
           mode=STANDARD,
           description="Tile shift in y-direction.")
protected int tileShiftY

Tile size in y-direction.

clusterNum

@Parameter(label="Number of feature clusters",
           required=true,
           dataIOOrder=10,
           direction=IN,
           mode=STANDARD,
           description="Number of feature clusters.")
protected int clusterNum

Number of clusters to be used in feature clustering.

doPCA

@Parameter(label="Do PCA in stage II?",
           required=true,
           dataIOOrder=11,
           direction=IN,
           mode=STANDARD,
           description="Enable/disable PCA prior to hierarchical clustering.")
protected boolean doPCA

Perform PCA in second stage?

maskDir

@Parameter(label="Cell boundary file folder",
           required=false,
           dataIOOrder=1,
           direction=IN,
           description="Cell mask directory.",
           mode=STANDARD)
protected de.unihalle.informatik.Alida.datatypes.ALDDirectoryString maskDir

(Optional) directory with (cell) boundaries.

stackedBarCharts

@Parameter(label="Resulting chart plots for each group",
           dataIOOrder=1,
           direction=OUT,
           description="Resulting chart plots.",
           mode=STANDARD)
protected Vector<org.jfree.chart.JFreeChart> stackedBarCharts

Resulting stacked bar plot of cluster distributions.

boxWhiskerCharts

@Parameter(label="Resulting box-whisker plot",
           dataIOOrder=1,
           direction=OUT,
           description="Resulting box-whisker plot.",
           mode=STANDARD)
protected Vector<org.jfree.chart.JFreeChart> boxWhiskerCharts

Box-whisker plot of the group-wise cluster distributions.

imageWidth

private transient int imageWidth

Width of the images taking first image as reference.

imageHeight

private transient int imageHeight

Height of the images taking first image as reference.

fileSep

private static String fileSep

File separator character, operating system dependent.

cellGroupNames

private transient TreeSet<String> cellGroupNames

List of group names, filled in clusterFeatures().

cellwiseDistros

private transient HashMap<String,double[]> cellwiseDistros

Map of cluster distributions per cell, filled in clusterFeatures().

cellwiseDistroKeys

private LinkedList<String> cellwiseDistroKeys

Sorted list of keys in cellwiseDistros once it is filled.

cellGroups

private transient Vector<HashMap<String,HashMap<String,Double>>> cellGroups

Group-wise cell distribution data, for each group the vector contains a hash map with the following structure: - the keys of the map are given by the cell IDs of the different cells (without basename which is equal to the group name) - the values are represented again as a hash map where the keys are given by the cluster IDs and the values by the relative frequency of the corresponding cluster in the cell Example: --------- cellGroups.get(0) = < 001-01, < [c1, vc1], ..., [c6, vc6] > > , ... < 005-07, < [c1, vc1], ..., [c6, vc6] > > cellGroups.get(1) = < 004-01, < [c1, vc1], ..., [c6, vc6] > > , ... < 008-01, < [c1, vc1], ..., [c6, vc6] > > The vector is filled in function clusterFeatures().

distroData

private transient double[][] distroData

Cluster distribution data, rows refer to clusters and columns to cells.

distroDataGroups

private transient String[] distroDataGroups

Group names linked to entries in distroData.

subspaceData

private transient double[][] subspaceData

Dimension-reduced cluster distribution data.

Constructor Detail

CytoskeletonAnalyzer2D

public CytoskeletonAnalyzer2D()
                       throws de.unihalle.informatik.Alida.exceptions.ALDOperatorException

Default constructor.

Throws:

de.unihalle.informatik.Alida.exceptions.ALDOperatorException - Thrown in case of failure.

Method Detail

operate

protected void operate()
                throws de.unihalle.informatik.Alida.exceptions.ALDOperatorException,
                       de.unihalle.informatik.Alida.exceptions.ALDProcessingDAGException

Specified by:

operate in class de.unihalle.informatik.Alida.operator.ALDOperator

Throws:

de.unihalle.informatik.Alida.exceptions.ALDOperatorException

de.unihalle.informatik.Alida.exceptions.ALDProcessingDAGException

clusterFeatures

private void clusterFeatures()
                      throws de.unihalle.informatik.Alida.exceptions.ALDOperatorException,
                             de.unihalle.informatik.Alida.exceptions.ALDProcessingDAGException

Performs the feature clustering via Weka's kMeans algorithm.

The method first reads the features from the given or formerly generated files, respectively, filters the feature vectors to exclude background tiles, and finally applies kMeans clustering. The feature files are expected to be located in the given feature directory.
The results are again saved to various files in the output directory:

• *-clusters.tif: visualization of cluster assignments
• *-clusters.ald: corresponding history file

Subsequently, the cluster distributions are analyzed in more detail, i.e., stacked bar plots and box-whisker plots are generated:

• *-clusterDistro.txt: cluster distributions per image and cell
• *-distributionChart.png: cell-wise distributions for each group
• AllGroupsClusterStatsChart.png: box-whisker plot of distributions
• AllImagesClusterStatistics.txt: cluster distributions for whole set

Throws:

de.unihalle.informatik.Alida.exceptions.ALDOperatorException - Thrown in case of failure.

de.unihalle.informatik.Alida.exceptions.ALDProcessingDAGException - Thrown in case of failure.

doPCA

private void doPCA()

Performs a Karhunen-Loeve transformation on the cluster distribution data.

The subspace dimension is chosen so that at least 95% of the data variance is represented within the resulting feature subspace. The result is saved to the file AllCellsPCASubspaceStats.txt.

calculateDistanceMatrices

private void calculateDistanceMatrices()

Calculates pairwise distance matrices and similarity network data.

Calculates pairwise Euclidean distances between all feature vectors and also between average vectors of all groups. In addition, for all groups a similarity network is constructed with each group forming a node and the edges representing the similarity between the groups. The bigger the weight of an edge the larger the similarity between two groups.

If PCA is enabled for stage II the distances are calculated from the subspace feature data as extracted by the PCA. If PCA is not enabled the cluster distribution data is used directly as base for the calculations.

The resulting distance matrices are saved to files with names AllCellsDistanceData.txt and AllGroupsDistanceData.txt in the root directory. The similarity network data is stored in a file named AllGroupsSimilarityNetworkData.txt.

calcFeatureFlagChanged

private void calcFeatureFlagChanged()

Callback routine to change parameters on change of flag for enable/disable feature calculation.

getDocumentation

public String getDocumentation()

Overrides:

getDocumentation in class de.unihalle.informatik.Alida.operator.ALDOperator