ParticleAnalyzer class (again)

classic Classic list List threaded Threaded
10 messages Options
Reply | Threaded
Open this post in threaded view
|

ParticleAnalyzer class (again)

Aryeh Weiss
Some years ago I asked how to process stacks using the ParticleAnalyzer
class. Here is the thread:
Imagej.1557.x6.nabble.com/ParticleAnalyzer-class-and-stacks-td5008596.html

However, I did not solve it, and I apparently let it drop. Now I am
trying again. Here is a code snippet that get a stack, but only works on
the first slice.

########################

# set up the parameters for the particle analyzer
     minSize = 1 # pixels;  We need to find a better basis for setting a
lower limit
     maxSize = width*height
     minCirc = 0.0
     maxCirc = 1.0
     options =
PA.SHOW_RESULTS+PA.EXCLUDE_EDGE_PARTICLES+PA.CLEAR_WORKSHEET+PA.ADD_TO_MANAGER+PA.IN_SITU_SHOW+PA.DOES_STACKS

     measurements = Measurements.ALL_STATS
     rt = ResultsTable()

     pa = PA(options, measurements, rt, minSize, maxSize, minCirc, maxCirc)
     print pa.processStack
     pa.analyze(lblImp, lblImp.getProcessor())

########################

A few thins to note:

1. I set DOES_STACK. The PA code indicates that if this is done, and the
image is a stack, then the protected variable processStack will be True.
2. I checked pa.processStack, and it is indeed True (pa is my instance
of the ParticleAnalyzer).
3. I checked the stack size of lblImp, and it is 11 (that is, greater
than 1)
4. I tried feeding pa.analyze an ImageStack - it did not want that
(cannot coerce to ImageProcessor)
5. I tried feeding pa.analyze the ImageProcessor, but that produces the
same result (only one mage processed). I tried giving it a StackProcessor,
     but again, that cannot be coerced to an ImageProcessor.

I ran the Particle Analyzer using the use interface to verify that it
runs on my image and can process the entire stack.

  I can loop through the ImageProcessors of the stack slices
(imageStack.getProcessor(n)), but it seems to me that the class is
capable of processing stacks, so my question is how?

Thanks in advance
--aryeh

--
Aryeh Weiss
Faculty of Engineering
Bar Ilan University
Ramat Gan 52900 Israel

Ph:  972-3-5317638
FAX: 972-3-7384051

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

ParticleAnalyzer class (again)

Aryeh Weiss
More PA strangeness. In the test script just below this text, I expect
to include Slice and Mode in my measurements.
I wrote the script as something minimal to see if I can get the PA class
to work on the entire stack , or to demonstrate that it does not.
Indeed it does not work on the entire stack, but in addition, Mode and
Slice do not appear in the ResultsTable rt. This is strange because in
my "large" script from which I took this code, I do get Mode and Slice.
Also, even in the  test script, rt.getColumnIndex("Mode") and
rt.getColumnIndex("Slice") appear
to return correct values for where those columns should be. When I run
Analyze Particle from the GUI , all works as expected, Mode is in column
3 (Area is column 0), and Slice is in column 26 (my script says 27). So
I am confused. Below is a script that should work on other
installations, so maybe some of you can tell me if you get similar
results. The output of the print statements is below the script.

Thanks in advance
--aryeh


#################################################
########### ParticleAnalyzer class test script##################

from ij import IJ,ImagePlus
from ij.plugin.filter import Analyzer, ParticleAnalyzer as PA
from ij.plugin.frame import RoiManager
from ij.measure import ResultsTable, Measurements
from ij.process import ImageProcessor

def testPA(imp):

     lblImp=imp.duplicate()

     width, height, nChannels, nSlices, nFrames = lblImp.getDimensions()

     IJ.setRawThreshold(lblImp, 50, 65535, "Red")
     lblImp.show()
     lblImp.setSlice(40)
# set up the parameters for the particle analyzer
     minSize = 1 # pixels;  We need to find a better basis for setting a
lower limit
     maxSize = width*height # This is a weak upper bound. If the ROI is
close to this, the segmentation is wrong
     minCirc = 0.0
     maxCirc = 1.0
     options =
PA.SHOW_RESULTS+PA.CLEAR_WORKSHEET+PA.ADD_TO_MANAGER+PA.IN_SITU_SHOW+PA.DOES_STACKS

     measurementsPA =
Measurements.ALL_STATS+Measurements.STACK_POSITION+Measurements.SLICE+Measurements.MODE
     rt = ResultsTable()

     pa = PA(options, measurementsPA, rt, minSize, maxSize, minCirc,
maxCirc)
     print "processStack = ",  pa.processStack
     print "StackSize = ", lblImp.getStackSize()

     pa.analyze(lblImp)
     rm = RoiManager().getInstance()

     rt.updateResults()
     rt.show("Results")
     print "Area column index: ", rt.getColumnIndex("Area")
     print "Mode column index: ", rt.getColumnIndex("Mode")
     print "Slice column index: ", rt.getColumnIndex("Slice")
     areas = rt.getColumn(rt.getColumnIndex("Area"))
     means = rt.getColumn(rt.getColumnIndex("Mode"))
     slices = rt.getColumn(rt.getColumnIndex("Slice"))
     print slices, "\n", means, "\n", areas

inputImp = IJ.openImage("http://imagej.nih.gov/ij/images/t1-head.zip")
testPA(inputImp)

########### end of ParticleAnalyzer class test script#############
################################################

##Script output:

processStack =  True
StackSize =  129
Area column index:  0
Mode column index:  3
Slice column index:  27
None
None
array('f', [35586.0703125, 2.250004529953003, 6.75001335144043,
2.250004529953003, 2.250004529953003, 2.250004529953003,
2.250004529953003, 4.500009059906006, 78.75016021728516,
2.250004529953003, 11.250022888183594])

## End of script output

On 27/08/2019 12:20, Aryeh Weiss wrote:

> Some years ago I asked how to process stacks using the
> ParticleAnalyzer class. Here is the thread:
> Imagej.1557.x6.nabble.com/ParticleAnalyzer-class-and-stacks-td5008596.html
>
>
> However, I did not solve it, and I apparently let it drop. Now I am
> trying again. Here is a code snippet that get a stack, but only works
> on the first slice.
>
> ########################
>
> # set up the parameters for the particle analyzer
>     minSize = 1 # pixels;  We need to find a better basis for setting
> a lower limit
>     maxSize = width*height
>     minCirc = 0.0
>     maxCirc = 1.0
>     options =
> PA.SHOW_RESULTS+PA.EXCLUDE_EDGE_PARTICLES+PA.CLEAR_WORKSHEET+PA.ADD_TO_MANAGER+PA.IN_SITU_SHOW+PA.DOES_STACKS
>
>     measurements = Measurements.ALL_STATS
>     rt = ResultsTable()
>
>     pa = PA(options, measurements, rt, minSize, maxSize, minCirc,
> maxCirc)
>     print pa.processStack
>     pa.analyze(lblImp, lblImp.getProcessor())
>
> ########################
>
> A few thins to note:
>
> 1. I set DOES_STACK. The PA code indicates that if this is done, and
> the image is a stack, then the protected variable processStack will be
> True.
> 2. I checked pa.processStack, and it is indeed True (pa is my instance
> of the ParticleAnalyzer).
> 3. I checked the stack size of lblImp, and it is 11 (that is, greater
> than 1)
> 4. I tried feeding pa.analyze an ImageStack - it did not want that
> (cannot coerce to ImageProcessor)
> 5. I tried feeding pa.analyze the ImageProcessor, but that produces
> the same result (only one mage processed). I tried giving it a
> StackProcessor,
>     but again, that cannot be coerced to an ImageProcessor.
>
> I ran the Particle Analyzer using the use interface to verify that it
> runs on my image and can process the entire stack.
>
>  I can loop through the ImageProcessors of the stack slices
> (imageStack.getProcessor(n)), but it seems to me that the class is
> capable of processing stacks, so my question is how?
>
> Thanks in advance
> --aryeh
>

--
Aryeh Weiss
Faculty of Engineering
Bar Ilan University
Ramat Gan 52900 Israel

Ph:  972-3-5317638
FAX: 972-3-7384051

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: ParticleAnalyzer class (again)

Michael Schmid-3
In reply to this post by Aryeh Weiss
Hi Aryeh,

when called from the GUI, a macro, or IJ.run, the ParticleAnalyzer
processes stacks via the usual method of PlugInFilters, where the
run(ip) method of the PlugInFilter is called for each slice of the stack
(this is done by the PlugInFIlterRunner). If you create a new
ParticleAnalyzer with the desired options and then call it analyze
method, it will work only on the current stack slice. See the source
code at:


https://github.com/imagej/imagej1/blob/master/ij/plugin/filter/ParticleAnalyzer.java#L429

You have to call it repeatedly for all slices (ImpagePlus.setSlice) if
you want to analyze the whole stack.

Michael
________________________________________________________________________

On 27/08/2019 11:20 AM, Aryeh Weiss wrote:
 > Some years ago I asked how to process stacks using the
ParticleAnalyzer class. Here is the thread:
 >
Imagej.1557.x6.nabble.com/ParticleAnalyzer-class-and-stacks-td5008596.html
 >
 > However, I did not solve it, and I apparently let it drop. Now I am
trying again. Here is a code snippet that get a stack, but only works on
the first slice.
 >
 > ########################
 >
 > # set up the parameters for the particle analyzer
 >      minSize = 1 # pixels;  We need to find a better basis for
setting a lower limit
 >      maxSize = width*height
 >      minCirc = 0.0
 >      maxCirc = 1.0
 >      options =
PA.SHOW_RESULTS+PA.EXCLUDE_EDGE_PARTICLES+PA.CLEAR_WORKSHEET+PA.ADD_TO_MANAGER+PA.IN_SITU_SHOW+PA.DOES_STACKS
 >
 >      measurements = Measurements.ALL_STATS
 >      rt = ResultsTable()
 >
 >      pa = PA(options, measurements, rt, minSize, maxSize, minCirc,
maxCirc)
 >      print pa.processStack
 >      pa.analyze(lblImp, lblImp.getProcessor())
 >
 > ########################
 >
 > A few thins to note:
 >
 > 1. I set DOES_STACK. The PA code indicates that if this is done, and
the image is a stack, then the protected variable processStack will be True.
 > 2. I checked pa.processStack, and it is indeed True (pa is my
instance of the ParticleAnalyzer).
 > 3. I checked the stack size of lblImp, and it is 11 (that is, greater
than 1)
 > 4. I tried feeding pa.analyze an ImageStack - it did not want that
(cannot coerce to ImageProcessor)
 > 5. I tried feeding pa.analyze the ImageProcessor, but that produces
the same result (only one mage processed). I tried giving it a
StackProcessor,
 >      but again, that cannot be coerced to an ImageProcessor.
 >
 > I ran the Particle Analyzer using the use interface to verify that it
runs on my image and can process the entire stack.
 >
 >   I can loop through the ImageProcessors of the stack slices
(imageStack.getProcessor(n)), but it seems to me that the class is
capable of processing stacks, so my question is how?
 >
 > Thanks in advance
 > --aryeh
 >

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: ParticleAnalyzer class (again)

Wayne Rasband-2
In reply to this post by Aryeh Weiss
> On Aug 28, 2019, at 4:18 AM, Aryeh Weiss <[hidden email]> wrote:
>
> More PA strangeness. In the test script just below this text, I expect to include Slice and Mode in my measurements.

The ParticleAnalyzer class does not process stacks. It is the IJ.run() method that runs the particle analyzer on all the images in a stack. An easy way to get the ROIs of all the particles in a stack is to run the particle analyzer using IJ.run(), have it create an overlay, retrieve the ROIs from the overlay and use ImagePlus.getStatistics (or ImagePlus.getAllStatistics) to get the desired measurements. The following JavaScript example does this:

-wayne

 img = IJ.openImage("http://imagej.nih.gov/ij/images/t1-head.zip")
 width = img.getWidth()
 height = img.getHeight()
 IJ.setRawThreshold(img, 50, 65535, "Red")
 maxSize = width*height
 options = "size=1-"+maxSize+" pixel show=Overlay clear stack"
 IJ.run(img, "Analyze Particles...", options);
 overlay = img.getOverlay()
 for (i=0; i<overlay.size(); i++) {
   roi = overlay.get(i)
   img.setRoi(roi)
   stats = img.getStatistics();
   slice = roi.getPosition();
   area = IJ.d2s(stats.area,2)
   mode = stats.mode
   print("roi="+i+", slice="+slice+", area="+area+", mode="+mode)
 }


> I wrote the script as something minimal to see if I can get the PA class to work on the entire stack , or to demonstrate that it does not.
> Indeed it does not work on the entire stack, but in addition, Mode and Slice do not appear in the ResultsTable rt. This is strange because in my "large" script from which I took this code, I do get Mode and Slice. Also, even in the  test script, rt.getColumnIndex("Mode") and rt.getColumnIndex("Slice") appear
> to return correct values for where those columns should be. When I run Analyze Particle from the GUI , all works as expected, Mode is in column 3 (Area is column 0), and Slice is in column 26 (my script says 27). So I am confused. Below is a script that should work on other installations, so maybe some of you can tell me if you get similar results. The output of the print statements is below the script.
>
> Thanks in advance
> --aryeh
>
>
> #################################################
> ########### ParticleAnalyzer class test script##################
>
> from ij import IJ,ImagePlus
> from ij.plugin.filter import Analyzer, ParticleAnalyzer as PA
> from ij.plugin.frame import RoiManager
> from ij.measure import ResultsTable, Measurements
> from ij.process import ImageProcessor
>
> def testPA(imp):
>
>     lblImp=imp.duplicate()
>
>     width, height, nChannels, nSlices, nFrames = lblImp.getDimensions()
>
>     IJ.setRawThreshold(lblImp, 50, 65535, "Red")
>     lblImp.show()
>     lblImp.setSlice(40)
> # set up the parameters for the particle analyzer
>     minSize = 1 # pixels;  We need to find a better basis for setting a lower limit
>     maxSize = width*height # This is a weak upper bound. If the ROI is close to this, the segmentation is wrong
>     minCirc = 0.0
>     maxCirc = 1.0
>     options = PA.SHOW_RESULTS+PA.CLEAR_WORKSHEET+PA.ADD_TO_MANAGER+PA.IN_SITU_SHOW+PA.DOES_STACKS
>
>     measurementsPA = Measurements.ALL_STATS+Measurements.STACK_POSITION+Measurements.SLICE+Measurements.MODE
>     rt = ResultsTable()
>
>     pa = PA(options, measurementsPA, rt, minSize, maxSize, minCirc, maxCirc)
>     print "processStack = ",  pa.processStack
>     print "StackSize = ", lblImp.getStackSize()
>
>     pa.analyze(lblImp)
>     rm = RoiManager().getInstance()
>
>     rt.updateResults()
>     rt.show("Results")
>     print "Area column index: ", rt.getColumnIndex("Area")
>     print "Mode column index: ", rt.getColumnIndex("Mode")
>     print "Slice column index: ", rt.getColumnIndex("Slice")
>     areas = rt.getColumn(rt.getColumnIndex("Area"))
>     means = rt.getColumn(rt.getColumnIndex("Mode"))
>     slices = rt.getColumn(rt.getColumnIndex("Slice"))
>     print slices, "\n", means, "\n", areas
>
> inputImp = IJ.openImage("http://imagej.nih.gov/ij/images/t1-head.zip")
> testPA(inputImp)
>
> ########### end of ParticleAnalyzer class test script#############
> ################################################
>
> ##Script output:
>
> processStack =  True
> StackSize =  129
> Area column index:  0
> Mode column index:  3
> Slice column index:  27
> None
> None
> array('f', [35586.0703125, 2.250004529953003, 6.75001335144043, 2.250004529953003, 2.250004529953003, 2.250004529953003, 2.250004529953003, 4.500009059906006, 78.75016021728516, 2.250004529953003, 11.250022888183594])
>
> ## End of script output
>
> On 27/08/2019 12:20, Aryeh Weiss wrote:
>> Some years ago I asked how to process stacks using the ParticleAnalyzer class. Here is the thread:
>> Imagej.1557.x6.nabble.com/ParticleAnalyzer-class-and-stacks-td5008596.html
>>
>> However, I did not solve it, and I apparently let it drop. Now I am trying again. Here is a code snippet that get a stack, but only works on the first slice.
>>
>> ########################
>>
>> # set up the parameters for the particle analyzer
>>     minSize = 1 # pixels;  We need to find a better basis for setting a lower limit
>>     maxSize = width*height
>>     minCirc = 0.0
>>     maxCirc = 1.0
>>     options = PA.SHOW_RESULTS+PA.EXCLUDE_EDGE_PARTICLES+PA.CLEAR_WORKSHEET+PA.ADD_TO_MANAGER+PA.IN_SITU_SHOW+PA.DOES_STACKS
>>
>>     measurements = Measurements.ALL_STATS
>>     rt = ResultsTable()
>>
>>     pa = PA(options, measurements, rt, minSize, maxSize, minCirc, maxCirc)
>>     print pa.processStack
>>     pa.analyze(lblImp, lblImp.getProcessor())
>>
>> ########################
>>
>> A few thins to note:
>>
>> 1. I set DOES_STACK. The PA code indicates that if this is done, and the image is a stack, then the protected variable processStack will be True.
>> 2. I checked pa.processStack, and it is indeed True (pa is my instance of the ParticleAnalyzer).
>> 3. I checked the stack size of lblImp, and it is 11 (that is, greater than 1)
>> 4. I tried feeding pa.analyze an ImageStack - it did not want that (cannot coerce to ImageProcessor)
>> 5. I tried feeding pa.analyze the ImageProcessor, but that produces the same result (only one mage processed). I tried giving it a StackProcessor,
>>     but again, that cannot be coerced to an ImageProcessor.
>>
>> I ran the Particle Analyzer using the use interface to verify that it runs on my image and can process the entire stack.
>>
>>  I can loop through the ImageProcessors of the stack slices (imageStack.getProcessor(n)), but it seems to me that the class is capable of processing stacks, so my question is how?
>>
>> Thanks in advance
>> --aryeh

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: ParticleAnalyzer class (again)

Aryeh Weiss
In reply to this post by Michael Schmid-3
Hi Michael,
Thank you for your reply.
I wonder what DOES_STACKS is for. I thought maybe it was to allow the PA
to work on the selected slice, but that happens also without DOES_STACKS.
This morning, i ran my test script, and it threw an error saying that pa
has not attribute process.Stack. Yesterday it did not say this, and the
source and docs seem to have it. Has the source just been changed?

Also,  why does the ResultsTable does not show columns for Slice and
Mode in my test script.

Note for the test script to run I now have to comment out
print pa.processStack

Best regards,
--aryeh

On 28/08/2019 20:58, Michael Schmid wrote:

> Hi Aryeh,
>
> when called from the GUI, a macro, or IJ.run, the ParticleAnalyzer
> processes stacks via the usual method of PlugInFilters, where the
> run(ip) method of the PlugInFilter is called for each slice of the
> stack (this is done by the PlugInFIlterRunner). If you create a new
> ParticleAnalyzer with the desired options and then call it analyze
> method, it will work only on the current stack slice. See the source
> code at:
>
>
> https://github.com/imagej/imagej1/blob/master/ij/plugin/filter/ParticleAnalyzer.java#L429 
>
>
> You have to call it repeatedly for all slices (ImpagePlus.setSlice) if
> you want to analyze the whole stack.
>
> Michael
> ________________________________________________________________________
>
> On 27/08/2019 11:20 AM, Aryeh Weiss wrote:
> > Some years ago I asked how to process stacks using the
> ParticleAnalyzer class. Here is the thread:
> >
> Imagej.1557.x6.nabble.com/ParticleAnalyzer-class-and-stacks-td5008596.html
> >
> > However, I did not solve it, and I apparently let it drop. Now I am
> trying again. Here is a code snippet that get a stack, but only works
> on the first slice.
> >
> > ########################
> >
> > # set up the parameters for the particle analyzer
> >      minSize = 1 # pixels;  We need to find a better basis for
> setting a lower limit
> >      maxSize = width*height
> >      minCirc = 0.0
> >      maxCirc = 1.0
> >      options =
> PA.SHOW_RESULTS+PA.EXCLUDE_EDGE_PARTICLES+PA.CLEAR_WORKSHEET+PA.ADD_TO_MANAGER+PA.IN_SITU_SHOW+PA.DOES_STACKS
> >
> >      measurements = Measurements.ALL_STATS
> >      rt = ResultsTable()
> >
> >      pa = PA(options, measurements, rt, minSize, maxSize, minCirc,
> maxCirc)
> >      print pa.processStack
> >      pa.analyze(lblImp, lblImp.getProcessor())
> >
> > ########################
> >
> > A few thins to note:
> >
> > 1. I set DOES_STACK. The PA code indicates that if this is done, and
> the image is a stack, then the protected variable processStack will be
> True.
> > 2. I checked pa.processStack, and it is indeed True (pa is my
> instance of the ParticleAnalyzer).
> > 3. I checked the stack size of lblImp, and it is 11 (that is,
> greater than 1)
> > 4. I tried feeding pa.analyze an ImageStack - it did not want that
> (cannot coerce to ImageProcessor)
> > 5. I tried feeding pa.analyze the ImageProcessor, but that produces
> the same result (only one mage processed). I tried giving it a
> StackProcessor,
> >      but again, that cannot be coerced to an ImageProcessor.
> >
> > I ran the Particle Analyzer using the use interface to verify that
> it runs on my image and can process the entire stack.
> >
> >   I can loop through the ImageProcessors of the stack slices
> (imageStack.getProcessor(n)), but it seems to me that the class is
> capable of processing stacks, so my question is how?
> >
> > Thanks in advance
> > --aryeh
> >
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
> .
>

--
Aryeh Weiss
Faculty of Engineering
Bar Ilan University
Ramat Gan 52900 Israel

Ph:  972-3-5317638
FAX: 972-3-7384051

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: ParticleAnalyzer class (again)

Aryeh Weiss
In reply to this post by Wayne Rasband-2
Hi Wayne and thank you for your reply.

On 28/08/2019 21:29, Wayne Rasband wrote:

>> On Aug 28, 2019, at 4:18 AM, Aryeh Weiss <[hidden email]> wrote:
>>
>> More PA strangeness. In the test script just below this text, I expect to include Slice and Mode in my measurements.
> The ParticleAnalyzer class does not process stacks. It is the IJ.run() method that runs the particle analyzer on all the images in a stack. An easy way to get the ROIs of all the particles in a stack is to run the particle analyzer using IJ.run(), have it create an overlay, retrieve the ROIs from the overlay and use ImagePlus.getStatistics (or ImagePlus.getAllStatistics) to get the desired measurements. The following JavaScript example does this:
>
> -wayne
>
>   img = IJ.openImage("http://imagej.nih.gov/ij/images/t1-head.zip")
>   width = img.getWidth()
>   height = img.getHeight()
>   IJ.setRawThreshold(img, 50, 65535, "Red")
>   maxSize = width*height
>   options = "size=1-"+maxSize+" pixel show=Overlay clear stack"
>   IJ.run(img, "Analyze Particles...", options);
>   overlay = img.getOverlay()
>   for (i=0; i<overlay.size(); i++) {
>     roi = overlay.get(i)
>     img.setRoi(roi)
>     stats = img.getStatistics();
>     slice = roi.getPosition();
>     area = IJ.d2s(stats.area,2)
>     mode = stats.mode
>     print("roi="+i+", slice="+slice+", area="+area+", mode="+mode)
>   }

I think that I can get all of the ROIs in each slice with one IJ.run
command:

IJ.run(imp, "Analyze Particles...", "size=1-Infinity pixel display clear
add in_situ stack");

This produces an RoiManager with all of the detected objects.

Best regards
--aryeh

>> I wrote the script as something minimal to see if I can get the PA class to work on the entire stack , or to demonstrate that it does not.
>> Indeed it does not work on the entire stack, but in addition, Mode and Slice do not appear in the ResultsTable rt. This is strange because in my "large" script from which I took this code, I do get Mode and Slice. Also, even in the  test script, rt.getColumnIndex("Mode") and rt.getColumnIndex("Slice") appear
>> to return correct values for where those columns should be. When I run Analyze Particle from the GUI , all works as expected, Mode is in column 3 (Area is column 0), and Slice is in column 26 (my script says 27). So I am confused. Below is a script that should work on other installations, so maybe some of you can tell me if you get similar results. The output of the print statements is below the script.
>>
>> Thanks in advance
>> --aryeh
>>
>>
>> #################################################
>> ########### ParticleAnalyzer class test script##################
>>
>> from ij import IJ,ImagePlus
>> from ij.plugin.filter import Analyzer, ParticleAnalyzer as PA
>> from ij.plugin.frame import RoiManager
>> from ij.measure import ResultsTable, Measurements
>> from ij.process import ImageProcessor
>>
>> def testPA(imp):
>>
>>      lblImp=imp.duplicate()
>>
>>      width, height, nChannels, nSlices, nFrames = lblImp.getDimensions()
>>
>>      IJ.setRawThreshold(lblImp, 50, 65535, "Red")
>>      lblImp.show()
>>      lblImp.setSlice(40)
>> # set up the parameters for the particle analyzer
>>      minSize = 1 # pixels;  We need to find a better basis for setting a lower limit
>>      maxSize = width*height # This is a weak upper bound. If the ROI is close to this, the segmentation is wrong
>>      minCirc = 0.0
>>      maxCirc = 1.0
>>      options = PA.SHOW_RESULTS+PA.CLEAR_WORKSHEET+PA.ADD_TO_MANAGER+PA.IN_SITU_SHOW+PA.DOES_STACKS
>>
>>      measurementsPA = Measurements.ALL_STATS+Measurements.STACK_POSITION+Measurements.SLICE+Measurements.MODE
>>      rt = ResultsTable()
>>
>>      pa = PA(options, measurementsPA, rt, minSize, maxSize, minCirc, maxCirc)
>>      print "processStack = ",  pa.processStack
>>      print "StackSize = ", lblImp.getStackSize()
>>
>>      pa.analyze(lblImp)
>>      rm = RoiManager().getInstance()
>>
>>      rt.updateResults()
>>      rt.show("Results")
>>      print "Area column index: ", rt.getColumnIndex("Area")
>>      print "Mode column index: ", rt.getColumnIndex("Mode")
>>      print "Slice column index: ", rt.getColumnIndex("Slice")
>>      areas = rt.getColumn(rt.getColumnIndex("Area"))
>>      means = rt.getColumn(rt.getColumnIndex("Mode"))
>>      slices = rt.getColumn(rt.getColumnIndex("Slice"))
>>      print slices, "\n", means, "\n", areas
>>
>> inputImp = IJ.openImage("http://imagej.nih.gov/ij/images/t1-head.zip")
>> testPA(inputImp)
>>
>> ########### end of ParticleAnalyzer class test script#############
>> ################################################
>>
>> ##Script output:
>>
>> processStack =  True
>> StackSize =  129
>> Area column index:  0
>> Mode column index:  3
>> Slice column index:  27
>> None
>> None
>> array('f', [35586.0703125, 2.250004529953003, 6.75001335144043, 2.250004529953003, 2.250004529953003, 2.250004529953003, 2.250004529953003, 4.500009059906006, 78.75016021728516, 2.250004529953003, 11.250022888183594])
>>
>> ## End of script output
>>
>> On 27/08/2019 12:20, Aryeh Weiss wrote:
>>> Some years ago I asked how to process stacks using the ParticleAnalyzer class. Here is the thread:
>>> Imagej.1557.x6.nabble.com/ParticleAnalyzer-class-and-stacks-td5008596.html
>>>
>>> However, I did not solve it, and I apparently let it drop. Now I am trying again. Here is a code snippet that get a stack, but only works on the first slice.
>>>
>>> ########################
>>>
>>> # set up the parameters for the particle analyzer
>>>      minSize = 1 # pixels;  We need to find a better basis for setting a lower limit
>>>      maxSize = width*height
>>>      minCirc = 0.0
>>>      maxCirc = 1.0
>>>      options = PA.SHOW_RESULTS+PA.EXCLUDE_EDGE_PARTICLES+PA.CLEAR_WORKSHEET+PA.ADD_TO_MANAGER+PA.IN_SITU_SHOW+PA.DOES_STACKS
>>>
>>>      measurements = Measurements.ALL_STATS
>>>      rt = ResultsTable()
>>>
>>>      pa = PA(options, measurements, rt, minSize, maxSize, minCirc, maxCirc)
>>>      print pa.processStack
>>>      pa.analyze(lblImp, lblImp.getProcessor())
>>>
>>> ########################
>>>
>>> A few thins to note:
>>>
>>> 1. I set DOES_STACK. The PA code indicates that if this is done, and the image is a stack, then the protected variable processStack will be True.
>>> 2. I checked pa.processStack, and it is indeed True (pa is my instance of the ParticleAnalyzer).
>>> 3. I checked the stack size of lblImp, and it is 11 (that is, greater than 1)
>>> 4. I tried feeding pa.analyze an ImageStack - it did not want that (cannot coerce to ImageProcessor)
>>> 5. I tried feeding pa.analyze the ImageProcessor, but that produces the same result (only one mage processed). I tried giving it a StackProcessor,
>>>      but again, that cannot be coerced to an ImageProcessor.
>>>
>>> I ran the Particle Analyzer using the use interface to verify that it runs on my image and can process the entire stack.
>>>
>>>   I can loop through the ImageProcessors of the stack slices (imageStack.getProcessor(n)), but it seems to me that the class is capable of processing stacks, so my question is how?
>>>
>>> Thanks in advance
>>> --aryeh
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

--
Aryeh Weiss
Faculty of Engineering
Bar Ilan University
Ramat Gan 52900 Israel

Ph:  972-3-5317638
FAX: 972-3-7384051

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: ParticleAnalyzer class (again)

Michael Schmid-3
In reply to this post by Aryeh Weiss
Hi Aryeh,

if the ParticleAnalyzer runs as a PlugInFilter (i.e., if it called from
the GUI, from a macro with 'run' or from Java/Javascript with 'IJ.run'),
the DOES_STACKS flag tells the PluginFilterRunner to call the
ParticleAnalyzers 'run' method for each stack slice.

Since there is no 'PARALLELIZE_STACKS' flag, this is done sequentially,
in a single thread.

The flags for PlugInFilters are specified in
https://github.com/imagej/imagej1/blob/master/ij/plugin/filter/PlugInFilter.java

There was no recent, relevant change of the ParticleAnalyzer that I
could see on github.

By the way, I think that the easiest for you would be to use
   IJ.run(imp, "Analyze Particles...", optionsString)

If you have an ImageStack (which is probably in an ImagePlus anyhow),
that would be easier than calling the ParticleAnalyzer for each slice,
and it is less likely that problems would arise in case the internal
workings of the ParticleAnalyzer should change in the future.

Michael
___________________________________________________________________________

On 29/08/2019 7:50 AM, Aryeh Weiss wrote:

> Hi Michael,
> Thank you for your reply.
> I wonder what DOES_STACKS is for. I thought maybe it was to allow the PA
> to work on the selected slice, but that happens also without DOES_STACKS.
> This morning, i ran my test script, and it threw an error saying that pa
> has not attribute process.Stack. Yesterday it did not say this, and the
> source and docs seem to have it. Has the source just been changed?
>
> Also,  why does the ResultsTable does not show columns for Slice and
> Mode in my test script.
>
> Note for the test script to run I now have to comment out
> print pa.processStack
>
> Best regards,
> --aryeh
>
> On 28/08/2019 20:58, Michael Schmid wrote:
>> Hi Aryeh,
>>
>> when called from the GUI, a macro, or IJ.run, the ParticleAnalyzer
>> processes stacks via the usual method of PlugInFilters, where the
>> run(ip) method of the PlugInFilter is called for each slice of the
>> stack (this is done by the PlugInFIlterRunner). If you create a new
>> ParticleAnalyzer with the desired options and then call it analyze
>> method, it will work only on the current stack slice. See the source
>> code at:
>>
>>
>> https://github.com/imagej/imagej1/blob/master/ij/plugin/filter/ParticleAnalyzer.java#L429 
>>
>>
>> You have to call it repeatedly for all slices (ImpagePlus.setSlice) if
>> you want to analyze the whole stack.
>>
>> Michael
>> ________________________________________________________________________
>>
>> On 27/08/2019 11:20 AM, Aryeh Weiss wrote:
>> > Some years ago I asked how to process stacks using the
>> ParticleAnalyzer class. Here is the thread:
>> >
>> Imagej.1557.x6.nabble.com/ParticleAnalyzer-class-and-stacks-td5008596.html
>>
>> >
>> > However, I did not solve it, and I apparently let it drop. Now I am
>> trying again. Here is a code snippet that get a stack, but only works
>> on the first slice.
>> >
>> > ########################
>> >
>> > # set up the parameters for the particle analyzer
>> >      minSize = 1 # pixels;  We need to find a better basis for
>> setting a lower limit
>> >      maxSize = width*height
>> >      minCirc = 0.0
>> >      maxCirc = 1.0
>> >      options =
>> PA.SHOW_RESULTS+PA.EXCLUDE_EDGE_PARTICLES+PA.CLEAR_WORKSHEET+PA.ADD_TO_MANAGER+PA.IN_SITU_SHOW+PA.DOES_STACKS
>>
>> >
>> >      measurements = Measurements.ALL_STATS
>> >      rt = ResultsTable()
>> >
>> >      pa = PA(options, measurements, rt, minSize, maxSize, minCirc,
>> maxCirc)
>> >      print pa.processStack
>> >      pa.analyze(lblImp, lblImp.getProcessor())
>> >
>> > ########################
>> >
>> > A few thins to note:
>> >
>> > 1. I set DOES_STACK. The PA code indicates that if this is done, and
>> the image is a stack, then the protected variable processStack will be
>> True.
>> > 2. I checked pa.processStack, and it is indeed True (pa is my
>> instance of the ParticleAnalyzer).
>> > 3. I checked the stack size of lblImp, and it is 11 (that is,
>> greater than 1)
>> > 4. I tried feeding pa.analyze an ImageStack - it did not want that
>> (cannot coerce to ImageProcessor)
>> > 5. I tried feeding pa.analyze the ImageProcessor, but that produces
>> the same result (only one mage processed). I tried giving it a
>> StackProcessor,
>> >      but again, that cannot be coerced to an ImageProcessor.
>> >
>> > I ran the Particle Analyzer using the use interface to verify that
>> it runs on my image and can process the entire stack.
>> >
>> >   I can loop through the ImageProcessors of the stack slices
>> (imageStack.getProcessor(n)), but it seems to me that the class is
>> capable of processing stacks, so my question is how?
>> >
>> > Thanks in advance
>> > --aryeh
>> >
>>
>> --
>> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>> .
>>
>

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: ParticleAnalyzer class (again)

Aryeh Weiss
Hi Michael.

Thank you for the explanation of that parameter. I still do not know why
the ResultsTable in my test script does not have the Slice and Mode columns.

I used IJ.run to run the particle analyzer, and it works as expected. My
motivation for using the ParticleAnalyzer class was that in one of my
scripts, when I ran the PA the script stopped, because the image
remained locked. This happened sometimes, not always, so I assume this
is also related to timing. Since I had in-situ show, when the RoiManager
tried to make its overlay, the GUI displayed a line saying that the
image was locked. I tried to reproduce this with a minimal script, where
I ran the PA many times in a loop, but that worked perfectly. When I
replaced the Ij.run() call with direct use of the PA class, this problem
went away.

Especially with methods that generate images, i try to use the API
classes directly, and not through IJ.run. If an IJ.run command generates
images, I  need to know how they will be named and grab them.  Also, if
they generate images, they will display them, which I do not always need
or want. If they display images, it maybe necessary to insert a loop
that waits until the window is available.

I should add that I started noticing these types of issues much more
when I moved to a Lenovo X1 Xtreme, Ubuntu 19.04 (POP_OS) using a 4K
display.

Best regards
--aryeh

On 29/08/2019 18:17, Michael Schmid wrote:

> Hi Aryeh,
>
> if the ParticleAnalyzer runs as a PlugInFilter (i.e., if it called
> from the GUI, from a macro with 'run' or from Java/Javascript with
> 'IJ.run'), the DOES_STACKS flag tells the PluginFilterRunner to call
> the ParticleAnalyzers 'run' method for each stack slice.
>
> Since there is no 'PARALLELIZE_STACKS' flag, this is done
> sequentially, in a single thread.
>
> The flags for PlugInFilters are specified in
> https://github.com/imagej/imagej1/blob/master/ij/plugin/filter/PlugInFilter.java 
>
>
> There was no recent, relevant change of the ParticleAnalyzer that I
> could see on github.
>
> By the way, I think that the easiest for you would be to use
>   IJ.run(imp, "Analyze Particles...", optionsString)
>
> If you have an ImageStack (which is probably in an ImagePlus anyhow),
> that would be easier than calling the ParticleAnalyzer for each slice,
> and it is less likely that problems would arise in case the internal
> workings of the ParticleAnalyzer should change in the future.
>
> Michael
> ___________________________________________________________________________
>
>
> On 29/08/2019 7:50 AM, Aryeh Weiss wrote:
>> Hi Michael,
>> Thank you for your reply.
>> I wonder what DOES_STACKS is for. I thought maybe it was to allow the
>> PA to work on the selected slice, but that happens also without
>> DOES_STACKS.
>> This morning, i ran my test script, and it threw an error saying that
>> pa has not attribute process.Stack. Yesterday it did not say this,
>> and the source and docs seem to have it. Has the source just been
>> changed?
>>
>> Also,  why does the ResultsTable does not show columns for Slice and
>> Mode in my test script.
>>
>> Note for the test script to run I now have to comment out
>> print pa.processStack
>>
>> Best regards,
>> --aryeh
>>
>> On 28/08/2019 20:58, Michael Schmid wrote:
>>> Hi Aryeh,
>>>
>>> when called from the GUI, a macro, or IJ.run, the ParticleAnalyzer
>>> processes stacks via the usual method of PlugInFilters, where the
>>> run(ip) method of the PlugInFilter is called for each slice of the
>>> stack (this is done by the PlugInFIlterRunner). If you create a new
>>> ParticleAnalyzer with the desired options and then call it analyze
>>> method, it will work only on the current stack slice. See the source
>>> code at:
>>>
>>>
>>> https://github.com/imagej/imagej1/blob/master/ij/plugin/filter/ParticleAnalyzer.java#L429 
>>>
>>>
>>> You have to call it repeatedly for all slices (ImpagePlus.setSlice)
>>> if you want to analyze the whole stack.
>>>
>>> Michael
>>> ________________________________________________________________________
>>>
>>>
>>> On 27/08/2019 11:20 AM, Aryeh Weiss wrote:
>>> > Some years ago I asked how to process stacks using the
>>> ParticleAnalyzer class. Here is the thread:
>>> >
>>> Imagej.1557.x6.nabble.com/ParticleAnalyzer-class-and-stacks-td5008596.html
>>>
>>> >
>>> > However, I did not solve it, and I apparently let it drop. Now I
>>> am trying again. Here is a code snippet that get a stack, but only
>>> works on the first slice.
>>> >
>>> > ########################
>>> >
>>> > # set up the parameters for the particle analyzer
>>> >      minSize = 1 # pixels;  We need to find a better basis for
>>> setting a lower limit
>>> >      maxSize = width*height
>>> >      minCirc = 0.0
>>> >      maxCirc = 1.0
>>> >      options =
>>> PA.SHOW_RESULTS+PA.EXCLUDE_EDGE_PARTICLES+PA.CLEAR_WORKSHEET+PA.ADD_TO_MANAGER+PA.IN_SITU_SHOW+PA.DOES_STACKS
>>>
>>> >
>>> >      measurements = Measurements.ALL_STATS
>>> >      rt = ResultsTable()
>>> >
>>> >      pa = PA(options, measurements, rt, minSize, maxSize, minCirc,
>>> maxCirc)
>>> >      print pa.processStack
>>> >      pa.analyze(lblImp, lblImp.getProcessor())
>>> >
>>> > ########################
>>> >
>>> > A few thins to note:
>>> >
>>> > 1. I set DOES_STACK. The PA code indicates that if this is done,
>>> and the image is a stack, then the protected variable processStack
>>> will be True.
>>> > 2. I checked pa.processStack, and it is indeed True (pa is my
>>> instance of the ParticleAnalyzer).
>>> > 3. I checked the stack size of lblImp, and it is 11 (that is,
>>> greater than 1)
>>> > 4. I tried feeding pa.analyze an ImageStack - it did not want that
>>> (cannot coerce to ImageProcessor)
>>> > 5. I tried feeding pa.analyze the ImageProcessor, but that
>>> produces the same result (only one mage processed). I tried giving
>>> it a StackProcessor,
>>> >      but again, that cannot be coerced to an ImageProcessor.
>>> >
>>> > I ran the Particle Analyzer using the use interface to verify that
>>> it runs on my image and can process the entire stack.
>>> >
>>> >   I can loop through the ImageProcessors of the stack slices
>>> (imageStack.getProcessor(n)), but it seems to me that the class is
>>> capable of processing stacks, so my question is how?
>>> >
>>> > Thanks in advance
>>> > --aryeh
>>> >
>>>
>>> --
>>> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>>> .
>>>
>>
>

--
Aryeh Weiss
Faculty of Engineering
Bar Ilan University
Ramat Gan 52900 Israel

Ph:  972-3-5317638
FAX: 972-3-7384051

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: ParticleAnalyzer class (again)

Michael Schmid-3
Hi Aryeh,

there was a bug (introduced by me when preparing for the non-blocking
dialogs for the filters - sorry) in one of the recent ImageJ versions,
causing the ParticleAnalyzer to create an improper "locked image" error
message. It should be fixed by now. In case you still see such a
problem, let me know, please!

Concerning images created by an IJ.run method:
If there is only one image that is created, it usually becomes the
foreground image (active image).
Also, the last image created is that with the most negative ImageID. So
if you have an IJ.run method creating only one image, you can use
WindowManager.getIDList() and determine the most negative imageID.
Then WindowManager.getImage(imageID) gives you a reference to the ImagePlus.
Of course, neither of these approaches is safe in case of parallel
operations due to multi-threading (but unfortunately, many ImageJ
operations are not multithrading-safe, so running things in parallel is
dangerous in much more respects)

Michael
______________________________________________________________


On 29/08/2019 7:26 PM, Aryeh Weiss wrote:

> Hi Michael.
>
> Thank you for the explanation of that parameter. I still do not know why
> the ResultsTable in my test script does not have the Slice and Mode
> columns.
>
> I used IJ.run to run the particle analyzer, and it works as expected. My
> motivation for using the ParticleAnalyzer class was that in one of my
> scripts, when I ran the PA the script stopped, because the image
> remained locked. This happened sometimes, not always, so I assume this
> is also related to timing. Since I had in-situ show, when the RoiManager
> tried to make its overlay, the GUI displayed a line saying that the
> image was locked. I tried to reproduce this with a minimal script, where
> I ran the PA many times in a loop, but that worked perfectly. When I
> replaced the Ij.run() call with direct use of the PA class, this problem
> went away.
>
> Especially with methods that generate images, i try to use the API
> classes directly, and not through IJ.run. If an IJ.run command generates
> images, I  need to know how they will be named and grab them.  Also, if
> they generate images, they will display them, which I do not always need
> or want. If they display images, it maybe necessary to insert a loop
> that waits until the window is available.
>
> I should add that I started noticing these types of issues much more
> when I moved to a Lenovo X1 Xtreme, Ubuntu 19.04 (POP_OS) using a 4K
> display.
>
> Best regards
> --aryeh
>
> On 29/08/2019 18:17, Michael Schmid wrote:
>> Hi Aryeh,
>>
>> if the ParticleAnalyzer runs as a PlugInFilter (i.e., if it called
>> from the GUI, from a macro with 'run' or from Java/Javascript with
>> 'IJ.run'), the DOES_STACKS flag tells the PluginFilterRunner to call
>> the ParticleAnalyzers 'run' method for each stack slice.
>>
>> Since there is no 'PARALLELIZE_STACKS' flag, this is done
>> sequentially, in a single thread.
>>
>> The flags for PlugInFilters are specified in
>> https://github.com/imagej/imagej1/blob/master/ij/plugin/filter/PlugInFilter.java 
>>
>>
>> There was no recent, relevant change of the ParticleAnalyzer that I
>> could see on github.
>>
>> By the way, I think that the easiest for you would be to use
>>   IJ.run(imp, "Analyze Particles...", optionsString)
>>
>> If you have an ImageStack (which is probably in an ImagePlus anyhow),
>> that would be easier than calling the ParticleAnalyzer for each slice,
>> and it is less likely that problems would arise in case the internal
>> workings of the ParticleAnalyzer should change in the future.
>>
>> Michael
>> ___________________________________________________________________________
>>
>>
>> On 29/08/2019 7:50 AM, Aryeh Weiss wrote:
>>> Hi Michael,
>>> Thank you for your reply.
>>> I wonder what DOES_STACKS is for. I thought maybe it was to allow the
>>> PA to work on the selected slice, but that happens also without
>>> DOES_STACKS.
>>> This morning, i ran my test script, and it threw an error saying that
>>> pa has not attribute process.Stack. Yesterday it did not say this,
>>> and the source and docs seem to have it. Has the source just been
>>> changed?
>>>
>>> Also,  why does the ResultsTable does not show columns for Slice and
>>> Mode in my test script.
>>>
>>> Note for the test script to run I now have to comment out
>>> print pa.processStack
>>>
>>> Best regards,
>>> --aryeh
>>>
>>> On 28/08/2019 20:58, Michael Schmid wrote:
>>>> Hi Aryeh,
>>>>
>>>> when called from the GUI, a macro, or IJ.run, the ParticleAnalyzer
>>>> processes stacks via the usual method of PlugInFilters, where the
>>>> run(ip) method of the PlugInFilter is called for each slice of the
>>>> stack (this is done by the PlugInFIlterRunner). If you create a new
>>>> ParticleAnalyzer with the desired options and then call it analyze
>>>> method, it will work only on the current stack slice. See the source
>>>> code at:
>>>>
>>>>
>>>> https://github.com/imagej/imagej1/blob/master/ij/plugin/filter/ParticleAnalyzer.java#L429 
>>>>
>>>>
>>>> You have to call it repeatedly for all slices (ImpagePlus.setSlice)
>>>> if you want to analyze the whole stack.
>>>>
>>>> Michael
>>>> ________________________________________________________________________
>>>>
>>>>
>>>> On 27/08/2019 11:20 AM, Aryeh Weiss wrote:
>>>> > Some years ago I asked how to process stacks using the
>>>> ParticleAnalyzer class. Here is the thread:
>>>> >
>>>> Imagej.1557.x6.nabble.com/ParticleAnalyzer-class-and-stacks-td5008596.html
>>>>
>>>> >
>>>> > However, I did not solve it, and I apparently let it drop. Now I
>>>> am trying again. Here is a code snippet that get a stack, but only
>>>> works on the first slice.
>>>> >
>>>> > ########################
>>>> >
>>>> > # set up the parameters for the particle analyzer
>>>> >      minSize = 1 # pixels;  We need to find a better basis for
>>>> setting a lower limit
>>>> >      maxSize = width*height
>>>> >      minCirc = 0.0
>>>> >      maxCirc = 1.0
>>>> >      options =
>>>> PA.SHOW_RESULTS+PA.EXCLUDE_EDGE_PARTICLES+PA.CLEAR_WORKSHEET+PA.ADD_TO_MANAGER+PA.IN_SITU_SHOW+PA.DOES_STACKS
>>>>
>>>> >
>>>> >      measurements = Measurements.ALL_STATS
>>>> >      rt = ResultsTable()
>>>> >
>>>> >      pa = PA(options, measurements, rt, minSize, maxSize, minCirc,
>>>> maxCirc)
>>>> >      print pa.processStack
>>>> >      pa.analyze(lblImp, lblImp.getProcessor())
>>>> >
>>>> > ########################
>>>> >
>>>> > A few thins to note:
>>>> >
>>>> > 1. I set DOES_STACK. The PA code indicates that if this is done,
>>>> and the image is a stack, then the protected variable processStack
>>>> will be True.
>>>> > 2. I checked pa.processStack, and it is indeed True (pa is my
>>>> instance of the ParticleAnalyzer).
>>>> > 3. I checked the stack size of lblImp, and it is 11 (that is,
>>>> greater than 1)
>>>> > 4. I tried feeding pa.analyze an ImageStack - it did not want that
>>>> (cannot coerce to ImageProcessor)
>>>> > 5. I tried feeding pa.analyze the ImageProcessor, but that
>>>> produces the same result (only one mage processed). I tried giving
>>>> it a StackProcessor,
>>>> >      but again, that cannot be coerced to an ImageProcessor.
>>>> >
>>>> > I ran the Particle Analyzer using the use interface to verify that
>>>> it runs on my image and can process the entire stack.
>>>> >
>>>> >   I can loop through the ImageProcessors of the stack slices
>>>> (imageStack.getProcessor(n)), but it seems to me that the class is
>>>> capable of processing stacks, so my question is how?
>>>> >
>>>> > Thanks in advance
>>>> > --aryeh
>>>> >
>>>>
>>>> --
>>>> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>>>> .
>>>>
>>>
>>
>

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: ParticleAnalyzer class (again)

Aryeh Weiss
Hi Michael

I am glad to hear that it is a known bug  that was corrected. I will let
you know if it recurs, but as I said, it was not consistently happening,
so I  do not know what triggered it.

Concerning created images - generally, the operations that create such
images create them with well defined names. Therefore, I wrote a
function called
getCreatedImage() that takes as its input the known title, and also a
second string which is the title I want (if I do not want "C1 - xxxxxx"
or whatever), and returns the imp, ip, and new image title. i will put
it here for the benefit of anyone following this thread who may find it
useful.

def getCreatedImage(inputName, nameStr):
     # This function returns the ImagePlus, ImageProcessor, and image
title associated
     # with an image that was opened with IJ.run(something that creates
an image)
     # inputName is thename of the created image, and nameStr is what it
will be renamed to.
     imp = WindowManager.getImage(inputName)
     imp.setTitle(nameStr)
     impTitle=imp.getTitle()
     ip = imp.getProcessor()
     return [imp, ip, impTitle]

Some operations create more than one image, in which case this function
will be called once for each created image.
Also, one must be careful not to have two images with the same name.

Best regards
--aryeh

On 29/08/2019 20:49, Michael Schmid wrote:

> Hi Aryeh,
>
> there was a bug (introduced by me when preparing for the non-blocking
> dialogs for the filters - sorry) in one of the recent ImageJ versions,
> causing the ParticleAnalyzer to create an improper "locked image"
> error message. It should be fixed by now. In case you still see such a
> problem, let me know, please!
>
> Concerning images created by an IJ.run method:
> If there is only one image that is created, it usually becomes the
> foreground image (active image).
> Also, the last image created is that with the most negative ImageID.
> So if you have an IJ.run method creating only one image, you can use
> WindowManager.getIDList() and determine the most negative imageID.
> Then WindowManager.getImage(imageID) gives you a reference to the
> ImagePlus.
> Of course, neither of these approaches is safe in case of parallel
> operations due to multi-threading (but unfortunately, many ImageJ
> operations are not multithrading-safe, so running things in parallel
> is dangerous in much more respects)
>
> Michael
> ______________________________________________________________
>
>
> On 29/08/2019 7:26 PM, Aryeh Weiss wrote:
>> Hi Michael.
>>
>> Thank you for the explanation of that parameter. I still do not know
>> why the ResultsTable in my test script does not have the Slice and
>> Mode columns.
>>
>> I used IJ.run to run the particle analyzer, and it works as expected.
>> My motivation for using the ParticleAnalyzer class was that in one of
>> my scripts, when I ran the PA the script stopped, because the image
>> remained locked. This happened sometimes, not always, so I assume
>> this is also related to timing. Since I had in-situ show, when the
>> RoiManager tried to make its overlay, the GUI displayed a line saying
>> that the image was locked. I tried to reproduce this with a minimal
>> script, where I ran the PA many times in a loop, but that worked
>> perfectly. When I replaced the Ij.run() call with direct use of the
>> PA class, this problem went away.
>>
>> Especially with methods that generate images, i try to use the API
>> classes directly, and not through IJ.run. If an IJ.run command
>> generates images, I  need to know how they will be named and grab
>> them.  Also, if they generate images, they will display them, which I
>> do not always need or want. If they display images, it maybe
>> necessary to insert a loop that waits until the window is available.
>>
>> I should add that I started noticing these types of issues much more
>> when I moved to a Lenovo X1 Xtreme, Ubuntu 19.04 (POP_OS) using a 4K
>> display.
>>
>> Best regards
>> --aryeh
>>
>> On 29/08/2019 18:17, Michael Schmid wrote:
>>> Hi Aryeh,
>>>
>>> if the ParticleAnalyzer runs as a PlugInFilter (i.e., if it called
>>> from the GUI, from a macro with 'run' or from Java/Javascript with
>>> 'IJ.run'), the DOES_STACKS flag tells the PluginFilterRunner to call
>>> the ParticleAnalyzers 'run' method for each stack slice.
>>>
>>> Since there is no 'PARALLELIZE_STACKS' flag, this is done
>>> sequentially, in a single thread.
>>>
>>> The flags for PlugInFilters are specified in
>>> https://github.com/imagej/imagej1/blob/master/ij/plugin/filter/PlugInFilter.java 
>>>
>>>
>>> There was no recent, relevant change of the ParticleAnalyzer that I
>>> could see on github.
>>>
>>> By the way, I think that the easiest for you would be to use
>>>   IJ.run(imp, "Analyze Particles...", optionsString)
>>>
>>> If you have an ImageStack (which is probably in an ImagePlus
>>> anyhow), that would be easier than calling the ParticleAnalyzer for
>>> each slice, and it is less likely that problems would arise in case
>>> the internal workings of the ParticleAnalyzer should change in the
>>> future.
>>>
>>> Michael
>>> ___________________________________________________________________________
>>>
>>>
>>> On 29/08/2019 7:50 AM, Aryeh Weiss wrote:
>>>> Hi Michael,
>>>> Thank you for your reply.
>>>> I wonder what DOES_STACKS is for. I thought maybe it was to allow
>>>> the PA to work on the selected slice, but that happens also without
>>>> DOES_STACKS.
>>>> This morning, i ran my test script, and it threw an error saying
>>>> that pa has not attribute process.Stack. Yesterday it did not say
>>>> this, and the source and docs seem to have it. Has the source just
>>>> been changed?
>>>>
>>>> Also,  why does the ResultsTable does not show columns for Slice
>>>> and Mode in my test script.
>>>>
>>>> Note for the test script to run I now have to comment out
>>>> print pa.processStack
>>>>
>>>> Best regards,
>>>> --aryeh
>>>>
>>>> On 28/08/2019 20:58, Michael Schmid wrote:
>>>>> Hi Aryeh,
>>>>>
>>>>> when called from the GUI, a macro, or IJ.run, the ParticleAnalyzer
>>>>> processes stacks via the usual method of PlugInFilters, where the
>>>>> run(ip) method of the PlugInFilter is called for each slice of the
>>>>> stack (this is done by the PlugInFIlterRunner). If you create a
>>>>> new ParticleAnalyzer with the desired options and then call it
>>>>> analyze method, it will work only on the current stack slice. See
>>>>> the source code at:
>>>>>
>>>>>
>>>>> https://github.com/imagej/imagej1/blob/master/ij/plugin/filter/ParticleAnalyzer.java#L429 
>>>>>
>>>>>
>>>>> You have to call it repeatedly for all slices
>>>>> (ImpagePlus.setSlice) if you want to analyze the whole stack.
>>>>>
>>>>> Michael
>>>>> ________________________________________________________________________
>>>>>
>>>>>
>>>>> On 27/08/2019 11:20 AM, Aryeh Weiss wrote:
>>>>> > Some years ago I asked how to process stacks using the
>>>>> ParticleAnalyzer class. Here is the thread:
>>>>> >
>>>>> Imagej.1557.x6.nabble.com/ParticleAnalyzer-class-and-stacks-td5008596.html
>>>>>
>>>>> >
>>>>> > However, I did not solve it, and I apparently let it drop. Now I
>>>>> am trying again. Here is a code snippet that get a stack, but only
>>>>> works on the first slice.
>>>>> >
>>>>> > ########################
>>>>> >
>>>>> > # set up the parameters for the particle analyzer
>>>>> >      minSize = 1 # pixels;  We need to find a better basis for
>>>>> setting a lower limit
>>>>> >      maxSize = width*height
>>>>> >      minCirc = 0.0
>>>>> >      maxCirc = 1.0
>>>>> >      options =
>>>>> PA.SHOW_RESULTS+PA.EXCLUDE_EDGE_PARTICLES+PA.CLEAR_WORKSHEET+PA.ADD_TO_MANAGER+PA.IN_SITU_SHOW+PA.DOES_STACKS
>>>>>
>>>>> >
>>>>> >      measurements = Measurements.ALL_STATS
>>>>> >      rt = ResultsTable()
>>>>> >
>>>>> >      pa = PA(options, measurements, rt, minSize, maxSize,
>>>>> minCirc, maxCirc)
>>>>> >      print pa.processStack
>>>>> >      pa.analyze(lblImp, lblImp.getProcessor())
>>>>> >
>>>>> > ########################
>>>>> >
>>>>> > A few thins to note:
>>>>> >
>>>>> > 1. I set DOES_STACK. The PA code indicates that if this is done,
>>>>> and the image is a stack, then the protected variable processStack
>>>>> will be True.
>>>>> > 2. I checked pa.processStack, and it is indeed True (pa is my
>>>>> instance of the ParticleAnalyzer).
>>>>> > 3. I checked the stack size of lblImp, and it is 11 (that is,
>>>>> greater than 1)
>>>>> > 4. I tried feeding pa.analyze an ImageStack - it did not want
>>>>> that (cannot coerce to ImageProcessor)
>>>>> > 5. I tried feeding pa.analyze the ImageProcessor, but that
>>>>> produces the same result (only one mage processed). I tried giving
>>>>> it a StackProcessor,
>>>>> >      but again, that cannot be coerced to an ImageProcessor.
>>>>> >
>>>>> > I ran the Particle Analyzer using the use interface to verify
>>>>> that it runs on my image and can process the entire stack.
>>>>> >
>>>>> >   I can loop through the ImageProcessors of the stack slices
>>>>> (imageStack.getProcessor(n)), but it seems to me that the class is
>>>>> capable of processing stacks, so my question is how?
>>>>> >
>>>>> > Thanks in advance
>>>>> > --aryeh
>>>>> >
>>>>>
>>>>> --
>>>>> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>>>>> .
>>>>>
>>>>
>>>
>>
>

--
Aryeh Weiss
Faculty of Engineering
Bar Ilan University
Ramat Gan 52900 Israel

Ph:  972-3-5317638
FAX: 972-3-7384051

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html