I notice that when I run StarDist (for 0.4.0) on Qupath 0.4.2 with the measurement options the displayed measurement keys for the segmented cells are different to the measurement keys displayed when running the standard cell detection. For example with StarDist the key is ordered '::' whereas the standard cell detection orders the measurement key ': .

See https://forum.image.sc/t/stardist-in-qupath-normalization-issue/38912/13 for a discussion.

Currently, if the number of classes expected by QuPath is higher than the classes provided by the model, QuPath will take the missing classes in the rays channels as the number of rays is computed from the number of expected classes. The warning for invalid classification size is then never triggered.

The operation for scaling is a subtraction instead of a multiplication.

See discussion at qupath/2022-qupath-hackathon#6

Bug report

Describe the bug
StarDist fails when run on an image with an incompatible tile size with an error such as

ERROR: RuntimeException: OpenCV(4.5.3) D:\a\javacpp-presets\javacpp-presets\opencv\cppbuild\windows-x86_64\opencv-4.5.3\modules\dnn\src\layers\concat_layer.cpp:102: error: (-201:Incorrect size of input array) Inconsistent shape for ConcatLayer in function 'cv::dnn::ConcatLayerImpl::getMemoryShapes'

When affecting smaller images, changing the tile size does not resolve the problem because the tiles are trimmed anyway based upon the image dimensions.

See https://forum.image.sc/t/error-running-stardist-on-qupath-v0-3-0-rc2-using-opencv-converted-model/56216 for details.

Expected behavior
No error. Image should automatically be padded to a compatible size immediately before prediction.

Version:

• v0.3.0-rc2 (possibly earlier versions as well)

Bug report

Describe the bug
Tile artifacts appear under some circumstances. See

• https://forum.image.sc/t/thoughts-on-0-3-2-rc2-with-stardist/56481
• https://forum.image.sc/t/qupath-v0-3-0-release-candidate-2/56185/4

To Reproduce
Not sure yet...

Expected behavior
No obvious tile artifacts.

Version

• v0.3.0-rc2

Additional context
Almost certainly due to the lack of padding since b09a16e
However, reinstating padding may not be entirely straightforward because of how tile requests are handled.

QuPath is not closing the loaded StarDist model when running on a GPU (QuPath built with CUDA). Each time a StarDist script is run, the model will be repeatedly loaded but never closed, leading to an out-of-memory exception when requesting memory greater than the available memory remaining in the VRAM.

Steps to reproduce outlined in: https://forum.image.sc/t/qupath-with-cuda-doesnt-release-gpu-memory-after-stardist-segmentation/57462

Hello!

When I try to use CLI to run an image processing task involving stardist I get the following error. It seems the workflow correctly loads stardist for the first call (as I am guessing from the "Added extension" line). Yet, on subsequent calls, I think something in the way batching works for scripts makes it lose track of where the Stardist installation lives.

"QuPath-0.3.2 (console).lnk" script -p="C:\Users\mashraf\Documents\QP_MAKEPROJ3\QuPathProject\project.qpproj" -s "C:\Users\mashraf\qp_batch_bash\da_stardist_v3_reg.groovy"

OpenJDK 64-Bit Server VM warning: Option --illegal-access is deprecated and will be removed in a future release.
15:38:46.089 [main] [INFO ] qupath.lib.common.ThreadTools - Setting parallelism to 15
15:38:46.256 [main] [INFO ] qupath.ScriptCommand - Setting tile cache size to 32600.00 MB (25.0% max memory)
15:38:46.366 [main] [INFO ] qupath.lib.gui.ExtensionClassLoader - Refreshing extensions in C:\Users\mashraf\Documents\qp_ext_test\extensions
15:38:46.367 [main] [INFO ] qupath.lib.gui.ExtensionClassLoader - Added extension: C:\Users\mashraf\Documents\qp_ext_test\extensions\qupath-extension-stardist-0.3.2.jar
15:38:46.381 [main] [INFO ] qupath.lib.scripting.QP - Initializing type adapters
15:38:47.761 [main] [INFO ] qupath.ScriptCommand - Running script for 2022_02_22__14_20__1786-Extended Depth of Focus-07-Scene-1-ScanRegion0_crop.czi - Scene #1
15:38:47.804 [main] [INFO ] q.l.i.s.o.OpenslideServerBuilder - OpenSlide version 3.4.1
15:38:47.926 [main] [INFO ] q.l.i.s.b.BioFormatsServerOptions - Setting max Bio-Formats readers to 16
C:\Users\mashraf\OneDrive - bluerocktx\Documents\qp_ext_test
15:38:56.479 [ForkJoinPool.commonPool-worker-25] [WARN ] qupath.opencv.ops.ImageOps - Normalization percentiles give the same value (0.0), scale will be Infinity
15:38:56.485 [ForkJoinPool.commonPool-worker-7] [WARN ] qupath.opencv.ops.ImageOps - Normalization percentiles give the same value (0.0), scale will be Infinity
15:38:56.635 [ForkJoinPool.commonPool-worker-13] [WARN ] qupath.opencv.ops.ImageOps - Normalization percentiles give the same value (0.0), scale will be Infinity
15:38:56.652 [ForkJoinPool.commonPool-worker-19] [WARN ] qupath.opencv.ops.ImageOps - Normalization percentiles give the same value (0.0), scale will be Infinity
15:38:56.861 [ForkJoinPool.commonPool-worker-23] [WARN ] qupath.opencv.ops.ImageOps - Normalization percentiles give the same value (0.0), scale will be Infinity
15:38:57.274 [ForkJoinPool.commonPool-worker-31] [WARN ] qupath.opencv.ops.ImageOps - Normalization percentiles give the same value (0.0), scale will be Infinity
15:38:59.955 [ForkJoinPool.commonPool-worker-1] [WARN ] qupath.opencv.ops.ImageOps - Normalization percentiles give the same value (0.0), scale will be Infinity
15:39:01.089 [ForkJoinPool.commonPool-worker-23] [WARN ] qupath.opencv.ops.ImageOps - Normalization percentiles give the same value (0.0), scale will be Infinity
15:39:04.608 [ForkJoinPool.commonPool-worker-9] [WARN ] qupath.opencv.ops.ImageOps - Normalization percentiles give the same value (0.0), scale will be Infinity
15:39:13.733 [main] [WARN ] qupath.lib.scripting.QP - Cannot run object classifier - no names provided!
15:39:13.745 [main] [INFO ] qupath.lib.io.PathIO - Writing object hierarchy with 1128 object(s)...
15:39:13.776 [main] [INFO ] qupath.lib.io.PathIO - Image data written in 0.04 seconds
15:39:13.819 [main] [INFO ] qupath.ScriptCommand - Running script for 2022_02_22__14_20__1786-Extended Depth of Focus-07-Scene-2-ScanRegion1_crop.czi - Scene #1
It looks like you have tried to import a class 'qupath.ext.stardist.StarDist' that doesn't exist!
MultipleCompilationErrorsException at line 11: startup failed:
Script2.groovy: 12: unable to resolve class qupath.ext.stardist.StarDist2D
@ line 12, column 1.
import qupath.ext.stardist.StarDist2D
^

1 error

org.codehaus.groovy.control.ErrorCollector.failIfErrors(ErrorCollector.java:295)
org.codehaus.groovy.control.CompilationUnit$ISourceUnitOperation.doPhaseOperation(CompilationUnit.java:914)
org.codehaus.groovy.control.CompilationUnit.compile(CompilationUnit.java:627)
groovy.lang.GroovyClassLoader.doParseClass(GroovyClassLoader.java:389)
groovy.lang.GroovyClassLoader.lambda$parseClass$3(GroovyClassLoader.java:332)
org.codehaus.groovy.runtime.memoize.StampedCommonCache.compute(StampedCommonCache.java:163)
org.codehaus.groovy.runtime.memoize.StampedCommonCache.getAndPut(StampedCommonCache.java:154)
groovy.lang.GroovyClassLoader.parseClass(GroovyClassLoader.java:330)
groovy.lang.GroovyClassLoader.parseClass(GroovyClassLoader.java:314)
groovy.lang.GroovyClassLoader.parseClass(GroovyClassLoader.java:257)
org.codehaus.groovy.jsr223.GroovyScriptEngineImpl.getScriptClass(GroovyScriptEngineImpl.java:336)
org.codehaus.groovy.jsr223.GroovyScriptEngineImpl.eval(GroovyScriptEngineImpl.java:153)
qupath.lib.gui.scripting.DefaultScriptEditor.executeScript(DefaultScriptEditor.java:982)
qupath.ScriptCommand.runScript(QuPath.java:432)
qupath.ScriptCommand.run(QuPath.java:319)
picocli.CommandLine.executeUserObject(CommandLine.java:1939)
picocli.CommandLine.access$1300(CommandLine.java:145)
picocli.CommandLine$RunLast.executeUserObjectOfLastSubcommandWithSameParent(CommandLine.java:2358)
picocli.CommandLine$RunLast.handle(CommandLine.java:2352)
picocli.CommandLine$RunLast.handle(CommandLine.java:2314)
picocli.CommandLine$AbstractParseResultHandler.execute(CommandLine.java:2179)
picocli.CommandLine$RunLast.execute(CommandLine.java:2316)
picocli.CommandLine.execute(CommandLine.java:2078)
qupath.QuPath.main(QuPath.java:187)
15:39:14.114 [main] [ERROR] qupath.ScriptCommand - Error running script for image: 2022_02_22__14_20__1786-Extended Depth of Focus-07-Scene-2-ScanRegion1_crop.czi - Scene #1
javax.script.ScriptException: org.codehaus.groovy.control.MultipleCompilationErrorsException: startup failed:
Script2.groovy: 12: unable to resolve class qupath.ext.stardist.StarDist2D
@ line 12, column 1.
import qupath.ext.stardist.StarDist2D
^

1 error

    at org.codehaus.groovy.jsr223.GroovyScriptEngineImpl.eval(GroovyScriptEngineImpl.java:158)
    at qupath.lib.gui.scripting.DefaultScriptEditor.executeScript(DefaultScriptEditor.java:982)
    at qupath.ScriptCommand.runScript(QuPath.java:432)
    at qupath.ScriptCommand.run(QuPath.java:319)
    at picocli.CommandLine.executeUserObject(CommandLine.java:1939)
    at picocli.CommandLine.access$1300(CommandLine.java:145)
    at picocli.CommandLine$RunLast.executeUserObjectOfLastSubcommandWithSameParent(CommandLine.java:2358)
    at picocli.CommandLine$RunLast.handle(CommandLine.java:2352)
    at picocli.CommandLine$RunLast.handle(CommandLine.java:2314)
    at picocli.CommandLine$AbstractParseResultHandler.execute(CommandLine.java:2179)
    at picocli.CommandLine$RunLast.execute(CommandLine.java:2316)
    at picocli.CommandLine.execute(CommandLine.java:2078)
    at qupath.QuPath.main(QuPath.java:187)

Caused by: org.codehaus.groovy.control.MultipleCompilationErrorsException: startup failed:
Script2.groovy: 12: unable to resolve class qupath.ext.stardist.StarDist2D
@ line 12, column 1.
import qupath.ext.stardist.StarDist2D

This error repeats a bunch of times, sometimes with the "Script2.groovy" section incrementing (although not in all cases). I should add that when I run the same script through the GUI it works just fine. Wondering if there is any way to force the batching to keep my extension path locked in? Open to any other solutions as well.

Thanks,

Mishal

Hello,

StarDist is running fine when I use a square annotation, but when it is no then I get an error.

Best regards,
Stefan

Version:
v0.3.0-rc2

MyScript:
import qupath.ext.stardist.StarDist2D

// Specify the model .pb file (you will need to change this!)
def pathModel = '/Qupath/QP_ET/StarDist_pre/dsb2018_heavy_augment.pb'

def stardist = StarDist2D.builder(pathModel)
        .threshold(0.5)              // Probability (detection) threshold
        .channels('Kerne-DAPI')            // Select detection channel
        .normalizePercentiles(1, 99) // Percentile normalization
        .pixelSize(0.5)              // Resolution for detection
        .cellExpansion(10.0)          // Approximate cells based upon nucleus expansion //Ursprüngl 5.0
        .cellConstrainScale(3.0)     // Constrain cell expansion using nucleus size // Urspr. 1.5
        .measureShape()              // Add shape measurements
        .measureIntensity()          // Add cell measurements (in all compartments)
        .includeProbability(true)    // Add probability as a measurement (enables later filtering)
        .doLog() // Logs, wenn alles läuft weglassen
        .build()

// Run detection for the selected objects
def imageData = getCurrentImageData()
def pathObjects = getSelectedObjects()
if (pathObjects.isEmpty()) {
    Dialogs.showErrorMessage("StarDist", "Please select a parent object!")
    return
}
stardist.detectObjects(imageData, pathObjects)
println 'Done!'

The error is:

INFO: Detecting nuclei for 266 tiles
WARN: Normalization percentiles give the same value (0.0), scale will be Infinity
WARN: Skipped 1 nucleus detection(s) due to error in resolving overlaps (0% of all skipped)
WARN: Normalization percentiles give the same value (0.0), scale will be Infinity
WARN: Skipped 1 nucleus detection(s) due to error in resolving overlaps (0% of all skipped)
WARN: Skipped 1 nucleus detection(s) due to error in resolving overlaps (0% of all skipped)
WARN: Normalization percentiles give the same value (0.0), scale will be Infinity
WARN: Skipped 1 nucleus detection(s) due to error in resolving overlaps (0% of all skipped)
WARN: Skipped 1 nucleus detection(s) due to error in resolving overlaps (0% of all skipped)
WARN: Normalization percentiles give the same value (0.0), scale will be Infinity
ERROR: RuntimeException: OpenCV(4.5.3) D:\a\javacpp-presets\javacpp-presets\opencv\cppbuild\windows-x86_64\opencv-4.5.3\modules\dnn\src\layers\concat_layer.cpp:102: error: (-201:Incorrect size of input array) Inconsistent shape for ConcatLayer in function 'cv::dnn::ConcatLayerImpl::getMemoryShapes'


ERROR: org.bytedeco.opencv.opencv_dnn.Net.forward(Native Method)
    qupath.opencv.dnn.OpenCVDnn$OpenCVNetFunction.predict(OpenCVDnn.java:690)
    qupath.opencv.dnn.OpenCVDnn$OpenCVNetFunction.predict(OpenCVDnn.java:704)
    qupath.opencv.dnn.DnnModel.convertAndPredict(DnnModel.java:100)
    qupath.ext.stardist.StarDist2D.detectObjectsForTile(StarDist2D.java:1020)
    qupath.ext.stardist.StarDist2D.lambda$detectObjects$5(StarDist2D.java:796)
    java.base/java.util.stream.ReferencePipeline$7$1.accept(Unknown Source)
    java.base/java.util.ArrayList$ArrayListSpliterator.forEachRemaining(Unknown Source)
    java.base/java.util.stream.AbstractPipeline.copyInto(Unknown Source)
    java.base/java.util.stream.AbstractPipeline.wrapAndCopyInto(Unknown Source)
    java.base/java.util.stream.ReduceOps$ReduceTask.doLeaf(Unknown Source)
    java.base/java.util.stream.ReduceOps$ReduceTask.doLeaf(Unknown Source)
    java.base/java.util.stream.AbstractTask.compute(Unknown Source)
    java.base/java.util.concurrent.CountedCompleter.exec(Unknown Source)
    java.base/java.util.concurrent.ForkJoinTask.doExec(Unknown Source)
    java.base/java.util.concurrent.ForkJoinPool$WorkQueue.topLevelExec(Unknown Source)
    java.base/java.util.concurrent.ForkJoinPool.scan(Unknown Source)
    java.base/java.util.concurrent.ForkJoinPool.runWorker(Unknown Source)
    java.base/java.util.concurrent.ForkJoinWorkerThread.run(Unknown Source)

See https://forum.image.sc/t/qupath-run-stardist-on-annotations-created-by-tissue-classifier/67215/2 for details.

Depending upon whether JTS internally optimizes things, the intersection calls like this one could be way more expensive that necessary.

We should also consider using OverlayNG - see:

• http://lin-ear-th-inking.blogspot.com/2020/06/jts-overlayng-noding-strategies.html
• http://lin-ear-th-inking.blogspot.com/2020/06/jts-overlayng-tolerant-topology.html