Skip to content

GitLab

The expiration time for new job artifacts in CI/CD pipelines is now 30 days (GitLab default). Previously generated artifacts in already completed jobs will not be affected by the change. The latest artifacts for all jobs in the latest successful pipelines will be kept. More information: https://gitlab.lrz.de/help/user/admin_area/settings/continuous_integration.html#default-artifacts-expiration

Commit 634d2db1 authored by Christian Schulte zu Berge's avatar Christian Schulte zu Berge
Browse files

Further work on TensorAnalyzer: added computation of various anisotropy measures, yet untested.

parent 1d6510ae
Hide whitespace changes
Inline Side-by-side
modules/tensor/processors/tensoranalyzer.cpp
View file @ 634d2db1
......@@ -32,7 +32,7 @@
#include <tbb/atomic.h>
#include "core/datastructures/imagedata.h"
#include "core/tools/stringutils.h"
namespace campvis {
......@@ -43,8 +43,32 @@ namespace campvis {
GenericOption<TensorAnalyzer::DegeneratedEvHandling>("shift", "Shift", TensorAnalyzer::SHIFT)
};
static const GenericOption<std::string> measurementOptions[15] = {
GenericOption<std::string>("Disabled", "Disabled"),
GenericOption<std::string>("EigenValue1", "Eigenvalue 1"),
GenericOption<std::string>("EigenValue2", "Eigenvalue 2"),
GenericOption<std::string>("EigenValue3", "Eigenvalue 3"),
GenericOption<std::string>("MainEigenvector", "Main Eigenvector"),
GenericOption<std::string>("VolumeRatio", "Volume Ratio"),
GenericOption<std::string>("FractionalAnisotropy", "Fractional Anisotropy"),
GenericOption<std::string>("RelativeAnisotropy", "Relative Anisotropy"),
GenericOption<std::string>("MeanDiffusivity", "Mean Diffusivity"),
GenericOption<std::string>("Trace", "Trace"),
GenericOption<std::string>("AxialDiffusivity", "Axial Diffusivity"),
GenericOption<std::string>("RadialDiffusivity", "Radial Diffusivity"),
GenericOption<std::string>("LinearAnisotropy", "Linear Anisotropy"),
GenericOption<std::string>("PlanarAnisotropy", "Planar Anisotropy"),
GenericOption<std::string>("Isotropy", "Isotropy")
};
const std::string TensorAnalyzer::loggerCat_ = "CAMPVis.modules.classification.TensorAnalyzer";
TensorAnalyzer::OutputPropertyPair::OutputPropertyPair(size_t index)
: _imageId("OutputId" + StringUtils::toString(index), "Output " + StringUtils::toString(index) + " Image", "TensorAnalyzer.output" + StringUtils::toString(index), DataNameProperty::WRITE)
, _imageType("OutputType" + StringUtils::toString(index), "Output " + StringUtils::toString(index) + " Image Type", measurementOptions, 15)
{}
TensorAnalyzer::TensorAnalyzer()
: AbstractProcessor()
, p_inputImage("InputImage", "Input Tensor Image", "tensors", DataNameProperty::READ, AbstractProcessor::INVALID_RESULT | EIGENSYSTEM_INVALID)
......@@ -52,6 +76,7 @@ namespace campvis {
, p_evecsImage("EvecsImage", "Output Eigenvectors Image", "eigenvectors", DataNameProperty::WRITE)
, p_degeneratedHandling("DegeneratedHandling", "Handling of Degenerated Tensors", handlingModes, 4)
, p_maskMixedTensors("MaskMixedTensors", "Mask Mixed Tensors", true)
, p_addOutputButton("AddOutputButton", "Add Output", AbstractProcessor::VALID)
, _eigenvalues(0)
, _eigenvectors(0)
{
......@@ -60,6 +85,10 @@ namespace campvis {
addProperty(&p_evecsImage);
addProperty(&p_degeneratedHandling);
addProperty(&p_maskMixedTensors);
addProperty(&p_addOutputButton);
addOutput();
p_addOutputButton.s_clicked.connect(this, &TensorAnalyzer::addOutput);
}
TensorAnalyzer::~TensorAnalyzer() {
......@@ -72,7 +101,9 @@ namespace campvis {
}
if (_eigenvalues.getData() != 0 && _eigenvectors.getData() != 0) {
for (size_t i = 0; i < p_outputProperties.size(); ++i) {
computeOutput(data, i);
}
}
else {
LERROR("Could not compute Eigensystem");
......@@ -195,4 +226,253 @@ namespace campvis {
validate(EIGENSYSTEM_INVALID);
}
void TensorAnalyzer::computeOutput(DataContainer& data, size_t index) {
// sanity check
if (index >= p_outputProperties.size()) {
LERROR("Index out of bounds while computing output #" << index);
return;
}
// gather eigensystem
const GenericImageRepresentationLocal<float, 3>* evalRep = 0;
const GenericImageRepresentationLocal<float, 9>* evecRep = 0;
if (_eigenvalues.getData() != 0 && _eigenvectors.getData() != 0) {
evalRep = static_cast<const ImageData*>(_eigenvalues.getData())->getRepresentation< GenericImageRepresentationLocal<float, 3> >(false);
evecRep = static_cast<const ImageData*>(_eigenvectors.getData())->getRepresentation< GenericImageRepresentationLocal<float, 9> >(false);
}
if (evalRep == 0 || evecRep == 0) {
LERROR("Could not compute output, no eigensystem present.");
return;
}
OutputPropertyPair* opp = p_outputProperties[index];
const std::string& type = opp->_imageType.getOptionValue();
if (type == "Disabled") {
return;
}
else if (type == "EigenValue1") {
ImageData* id = new ImageData(evalRep->getDimensionality(), evalRep->getSize(), 1);
GenericImageRepresentationLocal<float, 1>* output = GenericImageRepresentationLocal<float, 1>::create(id, 0);
tbb::parallel_for(tbb::blocked_range<size_t>(0, id->getNumElements()), [&] (const tbb::blocked_range<size_t>& range) {
for (size_t i = range.begin(); i != range.end(); ++i) {
const tgt::vec3& vals = evalRep->getElement(i);
float ev = vals.x;
if (ev == 0.0f || tgt::isNaN(ev))
output->setElement(i, 0.0f);
else
output->setElement(i, ev);
}
});
data.addData(opp->_imageId.getValue(), id);
}
else if (type == "EigenValue2") {
ImageData* id = new ImageData(evalRep->getDimensionality(), evalRep->getSize(), 1);
GenericImageRepresentationLocal<float, 1>* output = GenericImageRepresentationLocal<float, 1>::create(id, 0);
tbb::parallel_for(tbb::blocked_range<size_t>(0, id->getNumElements()), [&] (const tbb::blocked_range<size_t>& range) {
for (size_t i = range.begin(); i != range.end(); ++i) {
const tgt::vec3& vals = evalRep->getElement(i);
float ev = vals.y;
if (ev == 0.0f || tgt::isNaN(ev))
output->setElement(i, 0.0f);
else
output->setElement(i, ev);
}
});
data.addData(opp->_imageId.getValue(), id);
}
else if (type == "EigenValue3") {
ImageData* id = new ImageData(evalRep->getDimensionality(), evalRep->getSize(), 1);
GenericImageRepresentationLocal<float, 1>* output = GenericImageRepresentationLocal<float, 1>::create(id, 0);
tbb::parallel_for(tbb::blocked_range<size_t>(0, id->getNumElements()), [&] (const tbb::blocked_range<size_t>& range) {
for (size_t i = range.begin(); i != range.end(); ++i) {
const tgt::vec3& vals = evalRep->getElement(i);
float ev = vals.z;
if (ev == 0.0f || tgt::isNaN(ev))
output->setElement(i, 0.0f);
else
output->setElement(i, ev);
}
});
data.addData(opp->_imageId.getValue(), id);
}
else if (type == "MainEigenvector") {
ImageData* id = new ImageData(evalRep->getDimensionality(), evalRep->getSize(), 3);
GenericImageRepresentationLocal<float, 3>* output = GenericImageRepresentationLocal<float, 3>::create(id, 0);
tbb::parallel_for(tbb::blocked_range<size_t>(0, id->getNumElements()), [&] (const tbb::blocked_range<size_t>& range) {
for (size_t i = range.begin(); i != range.end(); ++i) {
output->setElement(i, evecRep->getElement(i)[0]);
}
});
data.addData(opp->_imageId.getValue(), id);
}
// = Anisotropy Measures: =========================================================================
else if (type == "VolumeRatio") {
ImageData* id = new ImageData(evalRep->getDimensionality(), evalRep->getSize(), 1);
GenericImageRepresentationLocal<float, 1>* output = GenericImageRepresentationLocal<float, 1>::create(id, 0);
tbb::parallel_for(tbb::blocked_range<size_t>(0, id->getNumElements()), [&] (const tbb::blocked_range<size_t>& range) {
for (size_t i = range.begin(); i != range.end(); ++i) {
const tgt::vec3& vals = evalRep->getElement(i);
if (vals == tgt::vec3::zero || tgt::isNaN(vals))
output->getElement(i) = 0;
else
output->getElement(i) = (vals.x * vals.y * vals.z) / pow((vals.x + vals.y + vals.z)/3.f, 3);
}
});
data.addData(opp->_imageId.getValue(), id);
}
else if (type == "FractionalAnisotropy") {
ImageData* id = new ImageData(evalRep->getDimensionality(), evalRep->getSize(), 1);
GenericImageRepresentationLocal<float, 1>* output = GenericImageRepresentationLocal<float, 1>::create(id, 0);
tbb::parallel_for(tbb::blocked_range<size_t>(0, id->getNumElements()), [&] (const tbb::blocked_range<size_t>& range) {
for (size_t i = range.begin(); i != range.end(); ++i) {
const tgt::vec3& vals = evalRep->getElement(i);
const float root = sqrt(.5f);
if (vals == tgt::vec3::zero || tgt::isNaN(vals))
output->getElement(i) = 0;
else
output->getElement(i) = root * sqrt((vals.x-vals.y)*(vals.x-vals.y) + (vals.y-vals.z)*(vals.y-vals.z) + (vals.z-vals.x)*(vals.z-vals.x)) / sqrt(vals.x*vals.x + vals.y*vals.y + vals.z*vals.z);
}
});
data.addData(opp->_imageId.getValue(), id);
}
else if (type == "RelativeAnisotropy") {
ImageData* id = new ImageData(evalRep->getDimensionality(), evalRep->getSize(), 1);
GenericImageRepresentationLocal<float, 1>* output = GenericImageRepresentationLocal<float, 1>::create(id, 0);
tbb::parallel_for(tbb::blocked_range<size_t>(0, id->getNumElements()), [&] (const tbb::blocked_range<size_t>& range) {
for (size_t i = range.begin(); i != range.end(); ++i) {
const tgt::vec3& vals = evalRep->getElement(i);
const float root = sqrt(.5f);
if (vals == tgt::vec3::zero || tgt::isNaN(vals))
output->getElement(i) = 0;
else
output->getElement(i) = root * sqrt((vals.x-vals.y)*(vals.x-vals.y) + (vals.y-vals.z)*(vals.y-vals.z) + (vals.z-vals.x)*(vals.z-vals.x)) / (vals.x + vals.y + vals.z);
}
});
data.addData(opp->_imageId.getValue(), id);
}
else if (type == "MeanDiffusivity") {
ImageData* id = new ImageData(evalRep->getDimensionality(), evalRep->getSize(), 1);
GenericImageRepresentationLocal<float, 1>* output = GenericImageRepresentationLocal<float, 1>::create(id, 0);
tbb::parallel_for(tbb::blocked_range<size_t>(0, id->getNumElements()), [&] (const tbb::blocked_range<size_t>& range) {
for (size_t i = range.begin(); i != range.end(); ++i) {
const tgt::vec3& vals = evalRep->getElement(i);
if (vals == tgt::vec3::zero || tgt::isNaN(vals))
output->getElement(i) = 0;
else
output->getElement(i) = (vals.x + vals.y + vals.z) / 3;
}
});
data.addData(opp->_imageId.getValue(), id);
}
else if (type == "Trace") {
ImageData* id = new ImageData(evalRep->getDimensionality(), evalRep->getSize(), 1);
GenericImageRepresentationLocal<float, 1>* output = GenericImageRepresentationLocal<float, 1>::create(id, 0);
tbb::parallel_for(tbb::blocked_range<size_t>(0, id->getNumElements()), [&] (const tbb::blocked_range<size_t>& range) {
for (size_t i = range.begin(); i != range.end(); ++i) {
const tgt::vec3& vals = evalRep->getElement(i);
if (vals == tgt::vec3::zero || tgt::isNaN(vals))
output->getElement(i) = 0;
else
output->getElement(i) = vals.x + vals.y + vals.z;
}
});
data.addData(opp->_imageId.getValue(), id);
}
else if (type == "AxialDiffusivity") {
ImageData* id = new ImageData(evalRep->getDimensionality(), evalRep->getSize(), 1);
GenericImageRepresentationLocal<float, 1>* output = GenericImageRepresentationLocal<float, 1>::create(id, 0);
tbb::parallel_for(tbb::blocked_range<size_t>(0, id->getNumElements()), [&] (const tbb::blocked_range<size_t>& range) {
for (size_t i = range.begin(); i != range.end(); ++i) {
const tgt::vec3& vals = evalRep->getElement(i);
if (vals == tgt::vec3::zero || tgt::isNaN(vals))
output->getElement(i) = 0;
else
output->getElement(i) = vals.x;
}
});
data.addData(opp->_imageId.getValue(), id);
}
else if (type == "RadialDiffusivity") {
ImageData* id = new ImageData(evalRep->getDimensionality(), evalRep->getSize(), 1);
GenericImageRepresentationLocal<float, 1>* output = GenericImageRepresentationLocal<float, 1>::create(id, 0);
tbb::parallel_for(tbb::blocked_range<size_t>(0, id->getNumElements()), [&] (const tbb::blocked_range<size_t>& range) {
for (size_t i = range.begin(); i != range.end(); ++i) {
const tgt::vec3& vals = evalRep->getElement(i);
if (vals == tgt::vec3::zero || tgt::isNaN(vals))
output->getElement(i) = 0;
else
output->getElement(i) = (vals.y + vals.z) / 2;
}
});
data.addData(opp->_imageId.getValue(), id);
}
else if (type == "LinearAnisotropy") {
ImageData* id = new ImageData(evalRep->getDimensionality(), evalRep->getSize(), 1);
GenericImageRepresentationLocal<float, 1>* output = GenericImageRepresentationLocal<float, 1>::create(id, 0);
tbb::parallel_for(tbb::blocked_range<size_t>(0, id->getNumElements()), [&] (const tbb::blocked_range<size_t>& range) {
for (size_t i = range.begin(); i != range.end(); ++i) {
const tgt::vec3& vals = evalRep->getElement(i);
if (vals == tgt::vec3::zero || tgt::isNaN(vals))
output->getElement(i) = 0;
else
output->getElement(i) = (vals.x - vals.y) / (vals.x + vals.y + vals.z);
}
});
data.addData(opp->_imageId.getValue(), id);
}
else if (type == "PlanarAnisotropy") {
ImageData* id = new ImageData(evalRep->getDimensionality(), evalRep->getSize(), 1);
GenericImageRepresentationLocal<float, 1>* output = GenericImageRepresentationLocal<float, 1>::create(id, 0);
tbb::parallel_for(tbb::blocked_range<size_t>(0, id->getNumElements()), [&] (const tbb::blocked_range<size_t>& range) {
for (size_t i = range.begin(); i != range.end(); ++i) {
const tgt::vec3& vals = evalRep->getElement(i);
if (vals == tgt::vec3::zero || tgt::isNaN(vals))
output->getElement(i) = 0;
else
output->getElement(i) = 2.f*(vals.y - vals.z) / (vals.x + vals.y + vals.z);
}
});
data.addData(opp->_imageId.getValue(), id);
}
else if (type == "Isotropy") {
ImageData* id = new ImageData(evalRep->getDimensionality(), evalRep->getSize(), 1);
GenericImageRepresentationLocal<float, 1>* output = GenericImageRepresentationLocal<float, 1>::create(id, 0);
tbb::parallel_for(tbb::blocked_range<size_t>(0, id->getNumElements()), [&] (const tbb::blocked_range<size_t>& range) {
for (size_t i = range.begin(); i != range.end(); ++i) {
const tgt::vec3& vals = evalRep->getElement(i);
if (vals == tgt::vec3::zero || tgt::isNaN(vals))
output->getElement(i) = 0;
else
output->getElement(i) = (3.f*vals.z) / (vals.x + vals.y + vals.z);
}
});
data.addData(opp->_imageId.getValue(), id);
}
}
void TensorAnalyzer::addOutput() {
OutputPropertyPair* opp = new OutputPropertyPair(p_outputProperties.size() + 1);
p_outputProperties.push_back(opp);
addProperty(&opp->_imageId);
addProperty(&opp->_imageType);
}
}
modules/tensor/processors/tensoranalyzer.h
View file @ 634d2db1
......@@ -29,6 +29,7 @@
#include "core/pipeline/visualizationprocessor.h"
#include "core/properties/datanameproperty.h"
#include "core/properties/buttonproperty.h"
#include "core/properties/genericproperty.h"
#include "core/properties/floatingpointproperty.h"
#include "core/properties/numericproperty.h"
......@@ -51,6 +52,14 @@ namespace campvis {
NONE, MASK, INVERT, SHIFT
};
// Pair of DataNameProperty for output image ID and OptionProperty for image type
struct OutputPropertyPair {
OutputPropertyPair(size_t index);;
DataNameProperty _imageId;
GenericOptionProperty<std::string> _imageType;
};
/**
* Constructs a new TensorAnalyzer Processor
**/
......@@ -79,6 +88,9 @@ namespace campvis {
GenericOptionProperty<DegeneratedEvHandling> p_degeneratedHandling; ///< Handling of degenerated tensors
BoolProperty p_maskMixedTensors;
ButtonProperty p_addOutputButton;
std::vector<OutputPropertyPair*> p_outputProperties;
protected:
/**
......@@ -88,6 +100,18 @@ namespace campvis {
*/
void computeEigensystem(DataContainer& data);
/**
* Computes the derived measurement for output number \a index.
* \param data DataContainer to store output image in.
* \param index Index of output to compute.
*/
void computeOutput(DataContainer& data, size_t index);
/**
* Adds another output for this processor (i.e. adds another OutputPropertyPair).
*/
void addOutput();
DataHandle _eigenvalues; ///< Current eigenvalues cached
DataHandle _eigenvectors; ///< Current eigenvectors cached
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment