/*-
* #%L
* Fiji's plugin for colocalization analysis.
* %%
* Copyright (C) 2009 - 2017 Fiji developers.
* %%
* This program is free software: you can redistribute it and/or modify
* it under the terms of the GNU General Public License as
* published by the Free Software Foundation, either version 3 of the
* License, or (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public
* License along with this program. If not, see
* <http://www.gnu.org/licenses/gpl-3.0.html>.
* #L%
*/
package sc.fiji.coloc.tests;
import static org.junit.Assert.assertEquals;
import net.imglib2.RandomAccessibleInterval;
import net.imglib2.TwinCursor;
import net.imglib2.type.logic.BitType;
import net.imglib2.type.numeric.integer.UnsignedByteType;
import net.imglib2.type.numeric.real.FloatType;
import net.imglib2.view.Views;
import org.junit.Test;
import sc.fiji.coloc.algorithms.MissingPreconditionException;
import sc.fiji.coloc.algorithms.PearsonsCorrelation;
import sc.fiji.coloc.algorithms.PearsonsCorrelation.Implementation;
import sc.fiji.coloc.gadgets.MaskFactory;
/**
* This class contains JUnit 4 test cases for the Pearson's correlation
* implementation.
*
* @author Dan White
* @author Tom Kazimiers
*/
public class PearsonsCorrelationTest extends ColocalisationTest {
/**
* Tests if the fast implementation of Pearson's correlation with two
* zero correlated images produce a Pearson's R value of about zero.
*/
@Test
public void fastPearsonsZeroCorrTest() throws MissingPreconditionException {
// create a twin value range cursor that iterates over all pixels of the input data
TwinCursor<UnsignedByteType> cursor = new TwinCursor<UnsignedByteType>(
zeroCorrelationImageCh1.randomAccess(),
zeroCorrelationImageCh2.randomAccess(),
Views.iterable(zeroCorrelationAlwaysTrueMask).localizingCursor());
// get the Pearson's value
double pearsonsR = PearsonsCorrelation.fastPearsons(cursor);
// check Pearsons R is close to zero
assertEquals(0.0, pearsonsR, 0.05);
}
/**
* Tests if the fast implementation of Pearson's correlation with two
* positive correlated images produce a Pearson's R value of about 0.75.
*/
@Test
public void fastPearsonsPositiveCorrTest() throws MissingPreconditionException {
// create a twin value range cursor that iterates over all pixels of the input data
TwinCursor<UnsignedByteType> cursor = new TwinCursor<UnsignedByteType>(
positiveCorrelationImageCh1.randomAccess(),
positiveCorrelationImageCh2.randomAccess(),
Views.iterable(positiveCorrelationAlwaysTrueMask).localizingCursor());
// get the Pearson's value
double pearsonsR = PearsonsCorrelation.fastPearsons(cursor);
// check Pearsons R is close to 0.75
assertEquals(0.75, pearsonsR, 0.01);
}
/**
* Tests if the classic implementation of Pearson's correlation with two
* zero correlated images produce a Pearson's R value of about zero.
*/
@Test
public void classicPearsonsZeroCorrTest() throws MissingPreconditionException {
// create a twin value range cursor that iterates over all pixels of the input data
TwinCursor<UnsignedByteType> cursor = new TwinCursor<UnsignedByteType>(
zeroCorrelationImageCh1.randomAccess(),
zeroCorrelationImageCh2.randomAccess(),
Views.iterable(zeroCorrelationAlwaysTrueMask).localizingCursor());
// get the Pearson's value
double pearsonsR = PearsonsCorrelation
.classicPearsons(cursor, zeroCorrelationImageCh1Mean, zeroCorrelationImageCh2Mean);
// check Pearsons R is close to zero
assertEquals(0.0, pearsonsR, 0.05);
}
/**
* Tests if the classic implementation of Pearson's correlation with two
* positive correlated images produce a Pearson's R value of about 0.75.
*/
@Test
public void classicPearsonsPositiveCorrTest() throws MissingPreconditionException {
// create a twin value range cursor that iterates over all pixels of the input data
TwinCursor<UnsignedByteType> cursor = new TwinCursor<UnsignedByteType>(
positiveCorrelationImageCh1.randomAccess(),
positiveCorrelationImageCh2.randomAccess(),
Views.iterable(positiveCorrelationAlwaysTrueMask).localizingCursor());
// get the Pearson's value
double pearsonsR = PearsonsCorrelation
.classicPearsons(cursor, positiveCorrelationImageCh1Mean, positiveCorrelationImageCh2Mean);
// check Pearsons R is close to 0.75
assertEquals(0.75, pearsonsR, 0.01);
}
/**
* Tests Pearson's correlation stays close to zero for image pairs with the same mean and spread
* of randomized pixel values around that mean.
*/
@Test
public void differentMeansTest() throws MissingPreconditionException {
final double initialMean = 0.2;
final double spread = 0.1;
final double[] sigma = new double[] {3.0, 3.0};
RandomAccessibleInterval<BitType> mask = MaskFactory.createMask(new long[] {512, 512}, true);
for (double mean = initialMean; mean < 1; mean += spread) {
RandomAccessibleInterval<FloatType> ch1 = TestImageAccessor.produceMeanBasedNoiseImage(new FloatType(),
512, 512, mean, spread, sigma, 0x01234567);
RandomAccessibleInterval<FloatType> ch2 = TestImageAccessor.produceMeanBasedNoiseImage(new FloatType(),
512, 512, mean, spread, sigma, 0x98765432);
// create a twin value range cursor that iterates over all pixels of the input data
TwinCursor<FloatType> cursor = new TwinCursor<FloatType>(ch1.randomAccess(),
ch2.randomAccess(), Views.iterable(mask).localizingCursor());
double resultFast = PearsonsCorrelation.fastPearsons(cursor);
assertEquals(0.0, resultFast, 0.1);
/* This test will throw Missing PreconsitionException, as the means are the same
* which causes a numerical problem in the classic implementation of Pearson's
* double resultClassic = PearsonsCorrelation.classicPearsons(cursor, mean, mean);
* assertTrue(Math.abs(resultClassic) < 0.1);
*/
}
}
/**
* The 1993 paper of Manders et. al about colocalization presents an own
* method and testing data for it. For that testing data there are
* Pearson colocalization numbers, too, and these get tested in this test.
* @throws MissingPreconditionException
*/
@Test
public void mandersPaperImagesTest() throws MissingPreconditionException {
PearsonsCorrelation<UnsignedByteType> pc =
new PearsonsCorrelation<UnsignedByteType>(Implementation.Classic);
double r;
// test A-A combination
r = pc.calculatePearsons(mandersA, mandersA, mandersAlwaysTrueMask);
assertEquals(1.0d, r, 0.01);
// test A-B combination
r = pc.calculatePearsons(mandersA, mandersB, mandersAlwaysTrueMask);
assertEquals(0.72d, r, 0.01);
// test A-C combination
r = pc.calculatePearsons(mandersA, mandersC, mandersAlwaysTrueMask);
assertEquals(0.44d, r, 0.01);
// test A-D combination
r = pc.calculatePearsons(mandersA, mandersD, mandersAlwaysTrueMask);
assertEquals(0.16d, r, 0.01);
// test A-E combination
r = pc.calculatePearsons(mandersA, mandersE, mandersAlwaysTrueMask);
assertEquals(-0.12d, r, 0.01);
// test A-F combination
r = pc.calculatePearsons(mandersA, mandersF, mandersAlwaysTrueMask);
assertEquals(0.22d, r, 0.01);
// test A-G combination
r = pc.calculatePearsons(mandersA, mandersG, mandersAlwaysTrueMask);
assertEquals(0.30d, r, 0.01);
// test A-H combination
r = pc.calculatePearsons(mandersA, mandersH, mandersAlwaysTrueMask);
assertEquals(0.48d, r, 0.01);
// test A-I combination
r = pc.calculatePearsons(mandersA, mandersI, mandersAlwaysTrueMask);
assertEquals(0.23d, r, 0.01);
}
}
