#/*########################################################################## # Copyright (C) 2004-2020 European Synchrotron Radiation Facility # # This file is part of the PyMca X-ray Fluorescence Toolkit developed at # the ESRF by the Software group. # # Permission is hereby granted, free of charge, to any person obtaining a copy # of this software and associated documentation files (the "Software"), to deal # in the Software without restriction, including without limitation the rights # to use, copy, modify, merge, publish, distribute, sublicense, and/or sell # copies of the Software, and to permit persons to whom the Software is # furnished to do so, subject to the following conditions: # # The above copyright notice and this permission notice shall be included in # all copies or substantial portions of the Software. # # THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR # IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, # FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE # AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER # LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, # OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN # THE SOFTWARE. # #############################################################################*/ """ This plugin opens a window allowing to configure and compute the principal component analysis. Each spectrum is considered an *observation*, and each channel is considered a *variable*. The user can configure following parameters: - PCA method (*Covariance, Expectation Max, Covariance Multiple Arrays*) - Number of Principal Components - Spectral Binning - Spectral Regions After the configuration dialog is validated, the eigenimages and the eigenvectors are computed and displayed in another window. """ # TODO: explain PCA methods and regions # TODO: provide a practical use case for a PCA. Isolating elements? __author__ = "V.A. Sole - ESRF Data Analysis" __contact__ = "sole@esrf.fr" __license__ = "MIT" __copyright__ = "European Synchrotron Radiation Facility, Grenoble, France" import numpy import logging from PyMca5 import StackPluginBase from PyMca5.PyMcaMath.mva import KMeansModule from PyMca5.PyMcaGui import CalculationThread from PyMca5.PyMcaGui.math.PCAWindow import PCAParametersDialog from PyMca5.PyMcaGui import StackPluginResultsWindow from PyMca5.PyMcaGui.pymca import RGBImageCalculator from PyMca5.PyMcaGui import PyMca_Icons qt = StackPluginResultsWindow.qt _logger = logging.getLogger(__name__) class PCAStackPlugin(StackPluginBase.StackPluginBase): def __init__(self, stackWindow, **kw): if _logger.getEffectiveLevel() == logging.DEBUG: StackPluginBase.pluginBaseLogger.setLevel(logging.DEBUG) StackPluginBase.StackPluginBase.__init__(self, stackWindow, **kw) self.methodDict = {'Calculate': [self.calculate, "Perform PCA", None], 'Show': [self._showWidget, "Show last results", PyMca_Icons.brushselect]} self.__methodKeys = ['Calculate', 'Show'] if 0 and KMeansModule.KMEANS: self.methodDict['KMeans'] = [self._showKMeansWidget, "KMeans", None] self.__methodKeys.append('KMeans') self.configurationWidget = None self.widget = None self._kMeansWidget = None self.thread = None def stackUpdated(self): _logger.debug("PCAStackPlugin.stackUpdated() called") self.configurationWidget = None self.widget = None self._kMeansWidget = None def selectionMaskUpdated(self): if self.widget is None: return if self.widget.isHidden(): if self._kMeansWidget is None: return elif self._kMeansWidget.isHidden(): return mask = self.getStackSelectionMask() self.widget.setSelectionMask(mask) if self._kMeansWidget: self._kMeansWidget.graphWidget.setSelectionMask(mask) def mySlot(self, ddict): _logger.debug("mySlot %s %s", ddict['event'], ddict.keys()) if ddict['event'] == "selectionMaskChanged": self.setStackSelectionMask(ddict['current']) elif ddict['event'] == "addImageClicked": self.addImage(ddict['image'], ddict['title']) elif ddict['event'] == "addAllClicked": for i in range(len(ddict["images"])): self.addImage(ddict['images'][i], ddict['titles'][i]) elif ddict['event'] == "removeImageClicked": self.removeImage(ddict['title']) elif ddict['event'] == "replaceImageClicked": self.replaceImage(ddict['image'], ddict['title']) elif ddict['event'] == "resetSelection": self.setStackSelectionMask(None) #Methods implemented by the plugin def getMethods(self): if self.widget is None: return [self.__methodKeys[0]] else: return self.__methodKeys def getMethodToolTip(self, name): return self.methodDict[name][1] def getMethodPixmap(self, name): return self.methodDict[name][2] def applyMethod(self, name): return self.methodDict[name][0]() #The specific part def calculate(self): if self.configurationWidget is None: stack = self.getStackDataObject() index = stack.info.get("McaIndex", -1) if index == len(stack.data.shape): index = -1 stack = None self.configurationWidget = PCAParametersDialog(None, regions=True, index=index) self._status = qt.QLabel(self.configurationWidget) self._status.setAlignment(qt.Qt.AlignHCenter) font = qt.QFont(self._status.font()) font.setBold(True) self._status.setFont(font) self._status.setText("Ready") self.configurationWidget.layout().addWidget(self._status) self.configurationWidget.setEnabled(True) activeCurve = self.getActiveCurve() if activeCurve is None: #I could get some defaults from the stack itslef raise ValueError("Please select an active curve") return x, spectrum, legend, info = activeCurve spectrumLength = max(spectrum.shape) oldValue = self.configurationWidget.nPC.value() self.configurationWidget.nPC.setMaximum(spectrumLength) self.configurationWidget.nPC.setValue(min(oldValue, spectrumLength)) binningOptions = [1] for number in [2, 3, 4, 5, 7, 9, 10, 11, 13, 15, 17, 19]: if (spectrumLength % number) == 0: binningOptions.append(number) # TODO: Should inform the configuration widget about the possibility # to encounter non-finite data? ddict = {'options': binningOptions, 'binning': 1, 'method': 0} self.configurationWidget.setParameters(ddict) y = spectrum self.configurationWidget.setSpectrum(x, y, legend=legend, info=info) self.configurationWidget.show() ret = self.configurationWidget.exec_() if ret: self._kMeansWidget = None self._executeFunctionAndParameters() def _executeFunctionAndParameters(self): self.widget = None self.configurationWidget.show() if _logger.getEffectiveLevel() == logging.DEBUG: self.thread = CalculationThread.CalculationThread(\ calculation_method=self.actualCalculation) self.thread.result = self.actualCalculation() self.threadFinished() else: self.thread = CalculationThread.CalculationThread(\ calculation_method=self.actualCalculation) self.thread.finished.connect(self.threadFinished) self.thread.start() message = "Please wait. PCA Calculation going on." CalculationThread.waitingMessageDialog(self.thread, message=message, parent=self.configurationWidget) def actualCalculation(self): pcaParameters = self.configurationWidget.getParameters() self._status.setText("Calculation going on") self.configurationWidget.setEnabled(False) #self.configurationWidget.close() self.__methodlabel = pcaParameters.get('methodlabel', "") function = pcaParameters['function'] pcaParameters['ncomponents'] = pcaParameters['npc'] # At some point I should make sure I get directly the # function and the parameters from the configuration widget del pcaParameters['npc'] del pcaParameters['method'] del pcaParameters['function'] del pcaParameters['methodlabel'] # binning = pcaParameters['binning'] # mask = pcaParameters['mask'] regions = pcaParameters['regions'] spectral_mask = pcaParameters['spectral_mask'] #print("regions = ", regions) #del pcaParameters['regions'] #del pcaParameters['spectral_mask'] #print("Regions and spectral mask not handled yet") if not self.isStackFinite(): # one has to check for NaNs in the used region(s) # for the time being only in the global image # spatial_mask = numpy.isfinite(image_data) spatial_mask = numpy.isfinite(self.getStackOriginalImage()) pcaParameters['mask'] = spatial_mask pcaParameters["legacy"] = False _logger.info("PCA function %s" % function.__name__) _logger.info("PCA parameters %s" % pcaParameters) if "Multiple" in self.__methodlabel: stackList = self.getStackDataObjectList() oldShapes = [] for stack in stackList: oldShapes.append(stack.data.shape) result = function(stackList, **pcaParameters) for i in range(len(stackList)): stackList[i].data.shape = oldShapes[i] return result else: stack = self.getStackDataObject() if isinstance(stack, numpy.ndarray): if stack.data.dtype not in [numpy.float, numpy.float32]: _logger.warning("WARNING: Non floating point data") text = "Calculation going on." text += " WARNING: Non floating point data." self._status.setText(text) oldShape = stack.data.shape result = function(stack, **pcaParameters) if stack.data.shape != oldShape: stack.data.shape = oldShape return result def threadFinished(self): result = self.thread.getResult() self.thread = None if type(result) == type((1,)): #if we receive a tuple there was an error if len(result): if result[0] == "Exception": self._status.setText("Ready after calculation error") self.configurationWidget.setEnabled(True) raise Exception(result[1], result[2]) return self._status.setText("Ready") curve = self.configurationWidget.getSpectrum(binned=True) if curve not in [None, []]: xValues = curve[0] else: xValues = None self.configurationWidget.setEnabled(True) self.configurationWidget.close() if hasattr(result, "keys"): # new way images = result["scores"] eigenValues = result["eigenvalues"] eigenVectors = result["eigenvectors"] variance = result.get("variance", None) if variance is not None: explainedVariance = [] for value in eigenValues: explainedVariance.append(100 * (value/variance)) else: variance = None images, eigenValues, eigenVectors = result methodlabel = self.__methodlabel imageNames = None vectorNames = None nimages = images.shape[0] imageNames = [] vectorNames = [] itmp = nimages if " ICA " in methodlabel: itmp = int(nimages / 2) for i in range(itmp): imageNames.append("ICAimage %02d" % i) vectorNames.append("ICAvector %02d" % i) if "Multiple" in methodlabel: xValues = None for i in range(itmp): imageNames.append("Eigenimage %02d" % i) vectorNames.append("Eigenvector %02d" % i) if variance is not None: vectorNames[-1] = "EV%02d Explained variance %.4f %%" % \ (i, explainedVariance[i]) self.widget = StackPluginResultsWindow.StackPluginResultsWindow(\ usetab=True) self.widget.buildAndConnectImageButtonBox(replace=True, multiple=True) qt = StackPluginResultsWindow.qt self.widget.sigMaskImageWidgetSignal.connect(self.mySlot) if xValues is not None: xValues = [xValues] * nimages self.widget.setStackPluginResults(images, spectra=eigenVectors, image_names=imageNames, xvalues=xValues, spectra_names=vectorNames) self._showWidget() def _showWidget(self): if self.widget is None: return #Show self.widget.show() self.widget.raise_() #update self.selectionMaskUpdated() def _showKMeansWidget(self): if self._kMeansWidget is None: self._kMeansWidget = RGBImageCalculator.RGBImageCalculator( \ math="kmeans", selection=True) #self._kMeansWidget = MaskImageWidget.MaskImageWidget() #labels = KMeansModule.label(view, k=int(min(nImages, 4))) #labels.shape = nRows, nColumns self._kMeansWidget.graphWidget.sigMaskImageWidgetSignal.connect( \ self.mySlot) # self._kMeansWidget.setImageData(labels) imageDict = {} for i in range(len(self.widget.imageList)): imageDict[self.widget.imageNames[i]] = \ {"image":self.widget.imageList[i]} self._kMeansWidget.imageList = list(imageDict.keys()) self._kMeansWidget.imageDict = imageDict #Show self._kMeansWidget.show() self._kMeansWidget.raise_() #update self.selectionMaskUpdated() MENU_TEXT = "PyMca PCA" def getStackPluginInstance(stackWindow, **kw): ob = PCAStackPlugin(stackWindow) return ob