def showDialogR(R, val):
""" Create dialog for input and save R values in settings
"""
gui = GenericDialog("Heterogeneous Z Correction")
msg = "Heterogeneous Z Correction calculates an optical model\n"\
"based on rational fluorescence intensity attenuation\n"\
"over depth, in the form of (ax+b)/(x+c). Please specify\n"\
"(a,b,c) for each region. You should provide a z-stack with\n"\
"unique pixel values for each region.\n"
gui.addMessage(msg)
for ii in range(len(val)):
gui.addNumericField(
"Region " + str(ii) + " (" + str(val[ii]) + "): a", R[ii][0], 2)
gui.addToSameRow()
gui.addNumericField("b", R[ii][1], 2)
gui.addToSameRow()
gui.addNumericField("c", R[ii][2], 2)
gui.addHelp(
r"https://github.com/alexandrebastien/heterogeneous-z-correction")
gui.showDialog()
if gui.wasOKed():
R = []
for ii in range(len(val)):
R = R + [
(gui.getNextNumber(), gui.getNextNumber(), gui.getNextNumber())
]
Prefs.set("HetZCorr.R", str(R))
return R
else:
return
def main():
metadatas = []
runagain = True
offsetx = 0
offsety = 0
interval = 50.0
IJ.run(imp, "RGB Color", "")
IJ.run("Colors...", "foreground=cyan background=white selection=yellow")
while runagain:
dialog = GenericDialog("Define 5x ROI parameters...")
dialog.enableYesNoCancel("Continue and quit",
"Continue and add another line")
dialog.addNumericField("Line scan position 1 X, Y: ", 0, 0)
dialog.addToSameRow()
dialog.addNumericField(", ", 0, 0)
dialog.addNumericField("Line scan position 2 X, Y: ", 0, 0)
dialog.addToSameRow()
dialog.addNumericField(", ", 0, 0)
dialog.addNumericField("Reference image crop origin X, Y: ", offsetx,
0)
dialog.addToSameRow()
dialog.addNumericField(", ", offsety, 0)
dialog.addNumericField("Frame interval (s): ", interval, 2)
dialog.showDialog()
p1x = dialog.getNextNumber()
p1y = dialog.getNextNumber()
p2x = dialog.getNextNumber()
p2y = dialog.getNextNumber()
offsetx = dialog.getNextNumber()
offsety = dialog.getNextNumber()
interval = dialog.getNextNumber()
linescan_overlay = LineScanOverlayMetadata(
(p1x, p1y), (p2x, p2y), (offsetx, offsety), interval)
metadatas.append(linescan_overlay)
roi = linescan_overlay.generateOverlayRoi()
imp.setRoi(roi)
IJ.run("Draw", "stack")
if dialog.wasOKed():
runagain = False
elif dialog.wasCanceled():
return
sd = SaveDialog("Save as AVI...",
os.path.splitext(imp.getTitle())[0],
"line scan overlay.avi")
if sd.getFileName is not None:
metadatas[0].save_to_json(
os.path.join(
sd.getDirectory(),
os.path.splitext(os.path.basename(sd.getFileName()))[0] +
"line scan metadata.json"))
if len(
metadatas
) > 1: # append: multiple writes=slow, but easiest way based on existing framework
for midx in range(1, len(metadatas)):
metadatas[midx].save_to_json(os.path.join(
sd.getDirectory(),
os.path.splitext(os.path.basename(sd.getFileName()))[0] +
"line scan metadata.json"),
append=True)
IJ.saveAs(
imp, "Tiff",
os.path.join(
sd.getDirectory(),
os.path.splitext(os.path.basename(sd.getFileName()))[0] +
" + line scan ROI, no timestamp.tif"))
IJ.run(
imp, "Label...",
"format=00:00:00 starting=0 interval=" + str(interval) +
" x=5 y=20 font=18 text=[] range=1-" + str(imp.getNFrames()))
IJ.run(
imp, "AVI... ", "compression=None frame=10 save=[" +
os.path.join(sd.getDirectory(),
(os.path.splitext(sd.getFileName())[0] +
" + line scan ROI.avi")) + "]")
IJ.saveAs(
imp, "Tiff",
os.path.join(
sd.getDirectory(),
os.path.splitext(os.path.basename(sd.getFileName()))[0] +
" + line scan ROI.tif"))
def run():
### Default arguments
two_sarcomere_size = 25 # determined by filter used.
rotation_angle = 0.0
### Get the image we'd like to work with.
# Don't need to ask where to save the intermediate image since it'll just be saved in the matchedmyo folder anyway
# may change this though. In case they want to keep that image around.
this_img = WindowManager.getCurrentImage()
if this_img == None:
ud = WaitForUserDialog(
"Please select the image you would like to analyze.")
ud.show()
this_img = WindowManager.getCurrentImage()
img_name = this_img.getTitle()
matchedmyo_path = "/home/AD/dfco222/scratchMarx/matchedmyo/" # this would be grabbed from the prompt
gd = GenericDialog("Preprocessing Options")
gd.addCheckbox("Automatic Resizing/Rotation", True)
gd.addCheckbox("CLAHE", True)
gd.addCheckbox("Normalization to Transverse Tubules", True)
gd.showDialog()
if gd.wasCanceled():
return
auto_resize_rotate = gd.getNextBoolean()
clahe = gd.getNextBoolean()
normalize = gd.getNextBoolean()
if auto_resize_rotate:
# Clear selection so it takes FFT of full image
rotation_angle = auto_resize_angle_measure(this_img,
two_sarcomere_size)
if clahe:
clahe_args = "blocksize={} histogram=256 maximum=3 mask=*None* fast_(less_accurate)".format(
two_sarcomere_size)
IJ.run(this_img, "Enhance Local Contrast (CLAHE)", clahe_args)
if normalize:
# Ask the user to select a subsection of the image that looks healthy-ish.
ud = WaitForUserDialog(
"Please select a subsection exhibiting a measure of healthy TT structure using the Rectangle tool.\n"
+ " Only a single cell or several are needed.\n\n" +
" Press 'OK' when finished.")
ud.show()
IJ.setTool("rectangle")
# Duplicate the selected subsection.
selection = this_img.crop()
IJ.run(selection, "Duplicate...", "title=subsection.tif")
# Grab the subsection image and rotate it.
selection = WindowManager.getImage("subsection.tif")
IJ.run(
selection, "Rotate...",
"angle={} grid=1 interpolation=Bicubic enlarge".format(
rotation_angle))
# Ask the user to select a bounding box that contains only tubules
# NOTE: Need to get rid of initial selection since it's the entire image and it's annoying to click out of
IJ.setTool("rectangle")
IJ.run(selection, "Select None", "")
ud = WaitForUserDialog(
"Select a subsection of the image that contains only tubules and no membrane."
)
ud.show()
# Grab the subsection ImagePlus
selection = WindowManager.getCurrentImage()
this_window = WindowManager.getActiveWindow()
selection_small = selection.crop()
IJ.run(selection, "Close", "")
# NOTE: May not actually display this depending on how the macros work
IJ.run(selection_small, "Duplicate...", "title=subsection_small.tif")
# Smooth the selection using the single TT filter.
# NOTE: It won't read in so we're just going to hard code it in since it's simple
tt_filt_row = "0 0 0 1 1 1 1 1 1 0 0 0 0\n"
tt_filt = ""
for i in range(21):
tt_filt += tt_filt_row
IJ.run("Convolve...", "text1=[" + tt_filt + "] normalize")
# Segment out the TTs from the 'gaps' using Gaussian Adaptive Thresholding.
selection_small = WindowManager.getImage("subsection_small.tif")
IJ.run(selection_small, "Duplicate...", "title=thresholded.tif")
threshed = WindowManager.getImage("thresholded.tif")
IJ.run(threshed, "Auto Local Threshold",
"method=Bernsen radius=7 parameter_1=1 parameter_2=0 white")
# Select the TTs from the thresholded image.
IJ.run(threshed, "Create Selection", "")
tt_selection = WindowManager.getImage("subsection_small.tif")
IJ.selectWindow("thresholded.tif")
IJ.selectWindow("subsection_small.tif")
IJ.run(tt_selection, "Restore Selection", "")
# Get TT intensity statistics.
stat_options = IS.MEAN | IS.MIN_MAX | IS.STD_DEV
stats = IS.getStatistics(tt_selection.getProcessor(), stat_options,
selection_small.getCalibration())
# Calculate pixel ceiling intensity value based on heuristic.
# TODO: Add a catch for data type overflow.
pixel_ceiling = stats.mean + 3 * stats.stdDev
print "px ceil:", pixel_ceiling
# Invert selection to get inter-sarcomeric gap intensity statistics.
IJ.run(tt_selection, "Make Inverse", "")
stat_options = IS.MEAN | IS.MIN_MAX | IS.STD_DEV
stats = IS.getStatistics(tt_selection.getProcessor(), stat_options,
selection_small.getCalibration())
# Calculate pixel floor intensity value based on heuristic.
pixel_floor = stats.mean - stats.stdDev
# TODO: Add a catch for data type underflow.
print "px floor:", pixel_floor
# Threshold original image based on these values.
IJ.selectWindow(this_img.getTitle())
IJ.run(this_img, "Select All", "")
IJ.setMinAndMax(pixel_floor, pixel_ceiling)
IJ.run(this_img, "Apply LUT", "")
## Ask if it is acceptable.
gd = GenericDialog("Acceptable?")
gd.addMessage(
"If the preprocessed image is acceptable for analysis, hit 'OK' to being analysis.\n"
+ " If the image is unacceptable or an error occurred, hit 'Cancel'")
gd.showDialog()
if gd.wasCanceled():
return
## Save the preprocessed image.
imp = IJ.getImage()
fs = FileSaver(imp)
img_save_dir = matchedmyo_path + "myoimages/" # actually get from user at some point
img_file_path = img_save_dir + img_name[:-4] + "_preprocessed.tif"
if os.path.exists(img_save_dir) and os.path.isdir(img_save_dir):
print "Saving image as:", img_file_path
if os.path.exists(img_file_path):
# use dialog box to ask if they want to overwrite
gd = GenericDialog("Overwrite?")
gd.addMessage(
"A file exists with the specified path, \"{}\". Would you like to overwrite it?"
.format(img_file_path))
gd.enableYesNoCancel()
gd.showDialog()
if gd.wasCanceled():
return
elif fs.saveAsTiff(img_file_path):
print "Preprocessed image saved successfully at:", '"' + img_file_path + '"'
else:
print "Folder does not exist or is not a folder!"
### Create the YAML file containing the parameters for classification
## Ask user for YAML input
gd = GenericDialog("YAML Input")
gd.addStringField("imageName", img_file_path, 50)
#gd.addStringField("maskName", "None")
gd.addStringField("outputParams_fileRoot", img_file_path[:-4], 50)
gd.addStringField("outputParams_fileType", "tif")
gd.addNumericField("outputParams_dpi", 300, 0)
gd.addCheckbox("outputParams_saveHitsArray", False)
gd.addStringField("outputParams_csvFile", matchedmyo_path + "results/")
gd.addCheckbox("TT Filtering", True)
gd.addToSameRow()
gd.addCheckbox("LT Filtering", True)
gd.addToSameRow()
gd.addCheckbox("TA Filtering", True)
gd.addNumericField("scopeResolutions_x", 5.0, 3)
gd.addToSameRow()
gd.addNumericField("scopeResolutions_y", 5.0, 3)
gd.addMessage("Enter in filter rotation angles separated by commas.")
gd.addStringField("", "-25, -20, -15, -10, -5, 0, 5, 10, 15, 20, 25", 50)
gd.addCheckbox("returnAngles", False)
gd.addCheckbox("returnPastedFilter", True)
gd.showDialog()
if gd.wasCanceled():
return
strings = [st.text for st in gd.getStringFields()]
#if strings[1] == "None" or "":
# strings[1] = None
nums = [float(num.text) for num in gd.getNumericFields()]
nums[0] = int(nums[0]) # Have to make sure the dpi variable is an integer
checks = [str(bool(boo.state)) for boo in gd.getCheckboxes()]
iter_argument = ','.join(
[str(float(it) - rotation_angle) for it in strings[4].split(',')])
string_block = """imageName: {0[0]}
outputParams:
fileRoot: {0[1]}
fileType: {0[2]}
dpi: {1[0]}
saveHitsArray: {2[0]}
csvFile: {0[3]}
preprocess: False
filterTypes:
TT: {2[1]}
LT: {2[2]}
TA: {2[3]}
scopeResolutions:
x: {1[1]}
y: {1[2]}
iters: [{3}]
returnAngles: {2[4]}
returnPastedFilter: {2[5]}""".format(strings, nums, checks, iter_argument)
im_title = this_img.getTitle()
with cd(matchedmyo_path):
yaml_file_path = "./YAML_files/" + im_title[:-4] + ".yml"
with open("./YAML_files/" + im_title[:-4] + ".yml", "w") as ym:
ym.write(string_block)
print "Wrote YAML file to:", matchedmyo_path + yaml_file_path[2:]
### Run the matchedmyo code on the preprocessed image
with cd(matchedmyo_path):
#os.chdir(matchedmyo_path)
#subprocess.call(["python3", matchedmyo_path+"matchedmyo.py", "fullValidation"])
subprocess.call(
["python3", "matchedmyo.py", "run", "--yamlFile", yaml_file_path])
