trackmate_cellpose.py

import sys

from fiji.plugin.trackmate import Model
from fiji.plugin.trackmate import Settings
from fiji.plugin.trackmate import TrackMate
from fiji.plugin.trackmate import Logger
from fiji.plugin.trackmate.io import TmXmlWriter
from fiji.plugin.trackmate.util import LogRecorder;
from fiji.plugin.trackmate.tracking.jaqaman import SparseLAPTrackerFactory, LAPUtils
from fiji.plugin.trackmate.util import TMUtils
from fiji.plugin.trackmate.visualization.hyperstack import HyperStackDisplayer
from fiji.plugin.trackmate import SelectionModel
from fiji.plugin.trackmate.cellpose import CellposeDetectorFactory
import fiji.plugin.trackmate.features.FeatureFilter as FeatureFilter
from fiji.plugin.trackmate.gui.displaysettings import DisplaySettings
from fiji.plugin.trackmate.gui.displaysettings import DisplaySettings
from fiji.plugin.trackmate.gui.displaysettings import DisplaySettingsIO
from fiji.plugin.trackmate.action import CaptureOverlayAction
from fiji.plugin.trackmate.cellpose.CellposeSettings import PretrainedModel

from ij import IJ


from datetime import datetime as dt

reload(sys)
sys.setdefaultencoding('utf-8')


# ------------------------------------------------------
# 	EDIT FILE PATHS BELOW.
# ------------------------------------------------------

# Shall we display the results each time?
show_output = True

# Channel to process? 
channel_to_process = 1

# Image files to analyse.
file_paths = []
file_paths.append('https://fiji.sc/samples/FakeTracks.tif' )


def run( image_file ):

    # Open image.
    imp  = IJ.openImage( image_file )
    cal = imp.getCalibration()

    # Logger -> content will be saved in the XML file.
    logger = LogRecorder( Logger.VOID_LOGGER )

    logger.log( 'TrackMate-Cellpose analysis script\n' )

    dt_string = dt.now().strftime("%d/%m/%Y %H:%M:%S")
    logger.log( dt_string + '\n\n' )

    #------------------------
    # Prepare settings object
    #------------------------

    settings = Settings(imp)
    setup = settings.toStringImageInfo() 

    # Configure Cellpose default detector.

    settings.detectorFactory = CellposeDetectorFactory()

    settings.detectorSettings['TARGET_CHANNEL'] = 1
    settings.detectorSettings['OPTIONAL_CHANNEL_2'] = 0
    settings.detectorSettings['CELLPOSE_PYTHON_FILEPATH'] = '/opt/anaconda3/envs/cellpose/bin/python'
    settings.detectorSettings['CELLPOSE_MODEL'] = PretrainedModel.CYTO2
    settings.detectorSettings['CELL_DIAMETER'] = 40.0
    settings.detectorSettings['USE_GPU'] = True
    settings.detectorSettings['SIMPLIFY_CONTOURS'] = False	


    # Configure tracker
    settings.trackerFactory = SparseLAPTrackerFactory()
    settings.trackerSettings = LAPUtils.getDefaultSegmentSettingsmap() #settings.trackerFactory.getDefaultSettings()#LAPUtils.getDefaultLAPSettingsMap()
    settings.trackerSettings[ 'LINKING_MAX_DISTANCE' ] 		= 5.0
    settings.trackerSettings[ 'GAP_CLOSING_MAX_DISTANCE' ]	= 5.0
    settings.trackerSettings[ 'MAX_FRAME_GAP' ]				= 2
    settings.initialSpotFilterValue = -1.

    # Analyzers 
    settings.addAllAnalyzers()

    # Add some filters for tracks/spots 

    # filter on track duration = keep tracks > 75% of total duration 
    duration_threshold = 75
    maxduration = (duration_threshold/100.0) * (imp.getNFrames() * cal.frameInterval)                
    filter1_track = FeatureFilter('TRACK_DURATION', maxduration, True)
    settings.addTrackFilter(filter1_track)

    # filter on spot = keep spots having radius > 1.6 um, and circularity > 0.7
    #filter1_spot = FeatureFilter('RADIUS', 1.6, True)
    #filter2_spot = FeatureFilter('CIRCULARITY', 0.7, True)
    #settings.addSpotFilter(filter1_spot)
    #settings.addSpotFilter(filter2_spot)

    print("Spot filters added = ", settings.getSpotFilters())
    print( "Track filters added = ", settings.getTrackFilters(), "\n")

    #-------------------
    # Instantiate plugin
    #-------------------

    trackmate = TrackMate( settings )
    trackmate.computeSpotFeatures( True )
    trackmate.computeTrackFeatures( True )
    trackmate.getModel().setLogger( logger )

    #--------
    # Process
    #--------

    ok = trackmate.checkInput()
    if not ok:
        print( str( trackmate.getErrorMessage() ) )
        return None
    ok = trackmate.process()
    if not ok:
        print( str( trackmate.getErrorMessage() ) )
        return None

    #----------------
    # Save results
    #----------------

    saveFile = TMUtils.proposeTrackMateSaveFile( settings, logger )

    writer = TmXmlWriter( saveFile, logger )
    writer.appendLog( logger.toString() )
    writer.appendModel( trackmate.getModel() )
    writer.appendSettings( trackmate.getSettings() )
    writer.writeToFile();
    print( "Results saved to: " + saveFile.toString() + '\n' )


    #----------------
    # Display results
    #----------------

    if show_output:
        model = trackmate.getModel()
        selectionModel = SelectionModel( model )
        ds = DisplaySettings()
        ds = DisplaySettingsIO.readUserDefault()
        ds.spotDisplayedAsRoi = True
        displayer =  HyperStackDisplayer( model, selectionModel, imp, ds )
        displayer.render()
        displayer.refresh()

    # capture overlay - RGB file
    image = trackmate.getSettings().imp
    capture = CaptureOverlayAction.capture(image, -1, imp.getNFrames(), logger)
    capture.setTitle("TracksOverlay")
    capture.show()

# ------------------------------------------------------

for file_path in file_paths:

    dt_string = dt.now().strftime("%d/%m/%Y %H:%M:%S")
    print( '\nRunning analysis on %s - %s' % ( file_path, dt_string ) )

    run( file_path )

print( 'Finished!' )