mitkCalculateGrayValueStatisticsTool.cpp

Go to the documentation of this file.

/*===================================================================

The Medical Imaging Interaction Toolkit (MITK)

Copyright (c) German Cancer Research Center,
Division of Medical and Biological Informatics.
All rights reserved.

This software is distributed WITHOUT ANY WARRANTY; without
even the implied warranty of MERCHANTABILITY or FITNESS FOR
A PARTICULAR PURPOSE.

See LICENSE.txt or http://www.mitk.org for details.

===================================================================*/

#include "mitkCalculateGrayValueStatisticsTool.h"

#include "mitkCalculateGrayValueStatisticsTool.xpm"

#include "mitkImageAccessByItk.h"
#include "mitkImageCast.h"
#include "mitkImageTimeSelector.h"
#include "mitkProgressBar.h"
#include "mitkStatusBar.h"
#include "mitkToolManager.h"

#include <itkCommand.h>

#include <itkHistogram.h>

namespace mitk
{
  MITK_TOOL_MACRO(MITKSEGMENTATION_EXPORT, CalculateGrayValueStatisticsTool, "Statistics tool");
}

mitk::CalculateGrayValueStatisticsTool::CalculateGrayValueStatisticsTool()
{
}

mitk::CalculateGrayValueStatisticsTool::~CalculateGrayValueStatisticsTool()
{
}

const char **mitk::CalculateGrayValueStatisticsTool::GetXPM() const
{
  return mitkCalculateGrayValueStatisticsTool_xpm;
}

const char *mitk::CalculateGrayValueStatisticsTool::GetName() const
{
  return "Statistics";
}

std::string mitk::CalculateGrayValueStatisticsTool::GetErrorMessage()
{
  return "No statistics generated for these segmentations:";
}

void mitk::CalculateGrayValueStatisticsTool::StartProcessingAllData()
{
  // clear/prepare report
  m_CompleteReport.str("");
}

bool mitk::CalculateGrayValueStatisticsTool::ProcessOneWorkingData(DataNode *node)
{
  if (node)
  {
    Image::Pointer image = dynamic_cast<Image *>(node->GetData());
    if (image.IsNull())
      return false;

    DataNode *referencenode = m_ToolManager->GetReferenceData(0);
    if (!referencenode)
      return false;

    try
    {
      ProgressBar::GetInstance()->AddStepsToDo(1);

      // add to report
      std::string nodename("structure");
      node->GetName(nodename);

      std::string message = "Calculating statistics for ";
      message += nodename;

      StatusBar::GetInstance()->DisplayText(message.c_str());

      Image::Pointer refImage = dynamic_cast<Image *>(referencenode->GetData());
      Image::Pointer image = dynamic_cast<Image *>(node->GetData());

      m_CompleteReport << "======== Gray value analysis of " << nodename << " ========\n";

      if (image.IsNotNull() && refImage.IsNotNull())
      {
        for (unsigned int timeStep = 0; timeStep < image->GetTimeSteps(); ++timeStep)
        {
          ImageTimeSelector::Pointer timeSelector = ImageTimeSelector::New();
          timeSelector->SetInput(refImage);
          timeSelector->SetTimeNr(timeStep);
          timeSelector->UpdateLargestPossibleRegion();
          Image::Pointer refImage3D = timeSelector->GetOutput();

          ImageTimeSelector::Pointer timeSelector2 = ImageTimeSelector::New();
          timeSelector2->SetInput(image);
          timeSelector2->SetTimeNr(timeStep);
          timeSelector2->UpdateLargestPossibleRegion();
          Image::Pointer image3D = timeSelector2->GetOutput();

          if (image3D.IsNotNull() && refImage3D.IsNotNull())
          {
            m_CompleteReport << "=== " << nodename << ", time step " << timeStep << " ===\n";
            AccessFixedDimensionByItk_2(refImage3D, ITKHistogramming, 3, image3D, m_CompleteReport);
          }
        }
      }

      m_CompleteReport << "======== End of analysis for " << nodename << " ===========\n\n\n";

      ProgressBar::GetInstance()->Progress();
    }
    catch (...)
    {
      return false;
    }
  }

  return true;
}

void mitk::CalculateGrayValueStatisticsTool::FinishProcessingAllData()
{
  SegmentationsProcessingTool::FinishProcessingAllData();

  // show/send results
  StatisticsCompleted.Send();
  // MITK_INFO << m_CompleteReport.str() << std::endl;
}

#define ROUND_P(x) ((int)((x) + 0.5))

template <typename TPixel, unsigned int VImageDimension>
void mitk::CalculateGrayValueStatisticsTool::CalculateMinMax(itk::Image<TPixel, VImageDimension> *referenceImage,
                                                             Image *segmentation,
                                                             TPixel &minimum,
                                                             TPixel &maximum)
{
  typedef itk::Image<TPixel, VImageDimension> ImageType;
  typedef itk::Image<Tool::DefaultSegmentationDataType, VImageDimension> SegmentationType;

  typename SegmentationType::Pointer segmentationItk;
  CastToItkImage(segmentation, segmentationItk);

  typename SegmentationType::RegionType segmentationRegion = segmentationItk->GetLargestPossibleRegion();

  segmentationRegion.Crop(referenceImage->GetLargestPossibleRegion());

  itk::ImageRegionConstIteratorWithIndex<SegmentationType> segmentationIterator(segmentationItk, segmentationRegion);
  itk::ImageRegionConstIteratorWithIndex<ImageType> referenceIterator(referenceImage, segmentationRegion);

  segmentationIterator.GoToBegin();
  referenceIterator.GoToBegin();

  minimum = std::numeric_limits<TPixel>::max();
  maximum = std::numeric_limits<TPixel>::min();

  while (!segmentationIterator.IsAtEnd())
  {
    itk::Point<float, 3> pt;
    segmentationItk->TransformIndexToPhysicalPoint(segmentationIterator.GetIndex(), pt);

    typename ImageType::IndexType ind;
    itk::ContinuousIndex<float, 3> contInd;
    if (referenceImage->TransformPhysicalPointToContinuousIndex(pt, contInd))
    {
      for (unsigned int i = 0; i < 3; ++i)
        ind[i] = ROUND_P(contInd[i]);

      referenceIterator.SetIndex(ind);

      if (segmentationIterator.Get() > 0)
      {
        if (referenceIterator.Get() < minimum)
          minimum = referenceIterator.Get();
        if (referenceIterator.Get() > maximum)
          maximum = referenceIterator.Get();
      }
    }

    ++segmentationIterator;
  }
}

template <typename TPixel, unsigned int VImageDimension>
void mitk::CalculateGrayValueStatisticsTool::ITKHistogramming(itk::Image<TPixel, VImageDimension> *referenceImage,
                                                              Image *segmentation,
                                                              std::stringstream &report)
{
  typedef itk::Image<TPixel, VImageDimension> ImageType;
  typedef itk::Image<Tool::DefaultSegmentationDataType, VImageDimension> SegmentationType;

  typename SegmentationType::Pointer segmentationItk;
  CastToItkImage(segmentation, segmentationItk);

  // generate histogram
  typename SegmentationType::RegionType segmentationRegion = segmentationItk->GetLargestPossibleRegion();

  segmentationRegion.Crop(referenceImage->GetLargestPossibleRegion());

  itk::ImageRegionConstIteratorWithIndex<SegmentationType> segmentationIterator(segmentationItk, segmentationRegion);
  itk::ImageRegionConstIteratorWithIndex<ImageType> referenceIterator(referenceImage, segmentationRegion);

  segmentationIterator.GoToBegin();
  referenceIterator.GoToBegin();

  m_ITKHistogram = HistogramType::New();

  TPixel minimum = std::numeric_limits<TPixel>::max();
  TPixel maximum = std::numeric_limits<TPixel>::min();

  CalculateMinMax(referenceImage, segmentation, minimum, maximum);

  // initialize the histogram to the range of the cropped region
  HistogramType::SizeType size;
#if defined(ITK_USE_REVIEW_STATISTICS)
  typedef typename HistogramType::SizeType::ValueType HSizeValueType;
#else
  typedef typename HistogramType::SizeType::SizeValueType HSizeValueType;
#endif
  size.Fill(static_cast<HSizeValueType>(maximum - minimum + 1));
  HistogramType::MeasurementVectorType lowerBound;
  HistogramType::MeasurementVectorType upperBound;
  lowerBound[0] = minimum;
  upperBound[0] = maximum;
  m_ITKHistogram->Initialize(size, lowerBound, upperBound);

  double mean(0.0);
  double sd(0.0);
  double voxelCount(0.0);

  // iterate through the cropped region add the values to the histogram
  while (!segmentationIterator.IsAtEnd())
  {
    itk::Point<float, 3> pt;
    segmentationItk->TransformIndexToPhysicalPoint(segmentationIterator.GetIndex(), pt);

    typename ImageType::IndexType ind;
    itk::ContinuousIndex<float, 3> contInd;
    if (referenceImage->TransformPhysicalPointToContinuousIndex(pt, contInd))
    {
      for (unsigned int i = 0; i < 3; ++i)
        ind[i] = ROUND_P(contInd[i]);

      referenceIterator.SetIndex(ind);

      if (segmentationIterator.Get() > 0)
      {
        HistogramType::MeasurementVectorType currentMeasurementVector;

        currentMeasurementVector[0] = static_cast<HistogramType::MeasurementType>(referenceIterator.Get());
        m_ITKHistogram->IncreaseFrequencyOfMeasurement(currentMeasurementVector, 1);

        mean = (mean * (static_cast<double>(voxelCount) /
                        static_cast<double>(voxelCount + 1))) // 3 points:   old center * 2/3 + currentPoint * 1/3;
               + static_cast<double>(referenceIterator.Get()) / static_cast<double>(voxelCount + 1);

        voxelCount += 1.0;
      }
    }

    ++segmentationIterator;
  }

  // second pass for SD
  segmentationIterator.GoToBegin();
  referenceIterator.GoToBegin();
  while (!segmentationIterator.IsAtEnd())
  {
    itk::Point<float, 3> pt;
    segmentationItk->TransformIndexToPhysicalPoint(segmentationIterator.GetIndex(), pt);

    typename ImageType::IndexType ind;
    itk::ContinuousIndex<float, 3> contInd;
    if (referenceImage->TransformPhysicalPointToContinuousIndex(pt, contInd))
    {
      for (unsigned int i = 0; i < 3; ++i)
        ind[i] = ROUND_P(contInd[i]);

      referenceIterator.SetIndex(ind);

      if (segmentationIterator.Get() > 0)
      {
        sd += ((static_cast<double>(referenceIterator.Get()) - mean) *
               (static_cast<double>(referenceIterator.Get()) - mean));
      }
    }

    ++segmentationIterator;
  }

  sd /= static_cast<double>(voxelCount - 1);
  sd = sqrt(sd);

  // generate quantiles
  TPixel histogramQuantileValues[5];
  histogramQuantileValues[0] = m_ITKHistogram->Quantile(0, 0.05);
  histogramQuantileValues[1] = m_ITKHistogram->Quantile(0, 0.25);
  histogramQuantileValues[2] = m_ITKHistogram->Quantile(0, 0.50);
  histogramQuantileValues[3] = m_ITKHistogram->Quantile(0, 0.75);
  histogramQuantileValues[4] = m_ITKHistogram->Quantile(0, 0.95);

  // report histogram values
  std::locale C("C");
  std::locale originalLocale = report.getloc();
  report.imbue(C);

  report << "         Minimum:" << minimum << "\n  5% quantile: " << histogramQuantileValues[0]
         << "\n 25% quantile: " << histogramQuantileValues[1] << "\n 50% quantile: " << histogramQuantileValues[2]
         << "\n 75% quantile: " << histogramQuantileValues[3] << "\n 95% quantile: " << histogramQuantileValues[4]
         << "\n      Maximum: " << maximum << "\n         Mean: " << mean << "\n           SD: " << sd << "\n";

  report.imbue(originalLocale);
}

std::string mitk::CalculateGrayValueStatisticsTool::GetReport() const
{
  return m_CompleteReport.str();
}

mitk::CalculateGrayValueStatisticsTool::HistogramType::ConstPointer
  mitk::CalculateGrayValueStatisticsTool::GetHistogram()
{
  return m_ITKHistogram.GetPointer();
}