diff --git a/src/bactos_functions.py b/src/bactos_functions.py
index cef843848a27560a8873091fee6d3a1b199648a4..2200a21ad10cc74a36a3bb6554d7d8cb31bb2734 100755
--- a/src/bactos_functions.py
+++ b/src/bactos_functions.py
@@ -9,15 +9,16 @@ import re
import json
import glob
import numpy as np
+from pathlib import Path
from tifffile import imread
-from skimage.feature import match_template
+from skimage.feature import match_template, match_descriptors, SIFT
from skimage.filters import gaussian, threshold_yen
from skimage.restoration import denoise_wavelet, estimate_sigma
from skimage.registration import phase_cross_correlation
from skimage.morphology import disk, white_tophat, opening
-from skimage.measure import label, regionprops, find_contours
+from skimage.measure import label, regionprops, find_contours, ransac
from skimage.segmentation import clear_border, expand_labels
-from skimage import transform
+from skimage import transform, exposure
from skimage.util import img_as_float
from scipy import ndimage as ndi
@@ -25,10 +26,14 @@ import dateutil
import h5py
import datetime
import os
+import io
from joblib import Parallel, delayed
from csbdeep.utils import normalize
from stardist.models import StarDist2D, Config2D
+from lib_bactos_hdf import load_image, load_mask, load_intensity_profile
+import matplotlib.pyplot as plt
+from rich.progress import track
def load_MM2_images(image_root_path, channel='*', position='*', time='*', z='*'):
@@ -285,11 +290,57 @@ def register_task(image, refimage, gaussian_size=0, up_factor=20, show_info=Fals
return dxdyt
+def register_task_sift(image, refimage, remove_cosmics=True, gaussian_size=0):
+ """
+ Use SIFT descriptor and RANSAC to obtain image shift
+ """
+
+ if remove_cosmics:
+ el = disk(2)
+ refimage = opening(refimage, el)
+ image = opening(image, el)
+
+ if gaussian_size > 0:
+ image = gaussian(image, gaussian_size)
+ refimage = gaussian(refimage, gaussian_size)
+
+ dstimg = exposure.rescale_intensity(image)
+ srcimg = exposure.rescale_intensity(refimage)
+
+ # SIFT detector
+ detect1 = SIFT()
+ detect2 = SIFT()
+
+ detect1.detect_and_extract(srcimg)
+ detect2.detect_and_extract(dstimg)
+
+ # Match detected points between the two images
+ try:
+ matched = match_descriptors(detect1.descriptors, detect2.descriptors,
+ max_ratio=0.6)
+ src = detect1.keypoints[matched[:, 0]]
+ dst = detect2.keypoints[matched[:, 1]]
+
+ # Use ransac to find the transform matrix
+ matrix, _ = ransac((src, dst), transform.EuclideanTransform,
+ min_samples=10,
+ residual_threshold=1)
+
+ # print(matrix.translation)
+ out = -matrix.translation
+ except:
+ out = (0, 0)
+
+ return out
+
+
def register_images(imgs_stack, gaussian_size = 0,
- register_filter=False, ref_filter=1,
+ register_filter=False, register_stack=True, ref_filter=1,
move_filter=0, show_info=False,
- up_factor=20, drift_filter=None, remove_cosmics=True,
- scale_correction=False):
+ up_factor=20, drift_filter=None,
+ remove_cosmics=True,
+ scale_correction=False,
+ drift_algo='cor'):
"""
Register images to correct drift in time and between filter
@@ -297,8 +348,20 @@ def register_images(imgs_stack, gaussian_size = 0,
remove_cosmics: True,
use an opening filter of disk size(2) to remove cosmics on images before registration
-
+ register_filter: boolean,
+ Do we need to register the other filter to the ref_filter
+
+ register_stack: boolean,
+ Do the registration of the whole stack
+
+ scale_correction: boolean,
+ Try to correct the scaling between filters (default False)
+
+ drift_algo: str in ['cor', 'sift'],
+ Define the algorithm to compute the drift between images.
+ 'cor': use crosscorrelation (the default)
+ 'sift': use SIFT point finding + RANSAC to find transform matrix
"""
# Do we need to do scale correction
@@ -311,47 +374,65 @@ def register_images(imgs_stack, gaussian_size = 0,
filters = [drift_filter]
# A first drift correction in time
- for pos in range(imgs_stack.shape[1]):
- dxdy = []
- for img_filter in filters:
- if show_info:
- print('stabilisation de la position pos %i' % pos)
- print('stabilisation du filtre %i' % img_filter)
-
- imgref = imgs_stack[0, pos, 0, 5:-5, 5:-5, img_filter]
-
- if remove_cosmics:
- el = disk(2)
- imgref = opening(imgref, el)
-
- if gaussian_size > 0:
- imgref = gaussian(imgref, gaussian_size)
-
- # Non parrallel version
- # dxdyt = [register_task(imgs_stack[t,pos,0,:,:,img_filter], imgref, up_factor=up_factor) for t in range(1, len(imgs_stack))]
- # Parralel version
- if remove_cosmics:
- dxdyt = Parallel(n_jobs=-1)(delayed(register_task)(opening(imgs_stack[t,pos,0, 5:-5, 5:-5, img_filter], el),
- imgref, up_factor=up_factor) for t in range(1, len(imgs_stack)))
- else:
- dxdyt = Parallel(n_jobs=-1)(delayed(register_task)(imgs_stack[t,pos,0, 5:-5, 5:-5, img_filter], imgref, up_factor=up_factor) for t in range(1, len(imgs_stack)))
- dxdy += dxdyt
-
- if drift_filter is None:
- for i, t in enumerate(range(1, len(imgs_stack))):
- imgs_stack[t, pos, 0,:,:,img_filter] = ndi.shift(imgs_stack[t,pos,0,:,:,img_filter], dxdyt[i])
-
- if drift_filter is not None:
- for img_filter in range(imgs_stack.shape[5]):
+ if register_stack:
+ print('Start stack registration !')
+ for pos in range(imgs_stack.shape[1]):
+ dxdy = []
+ for img_filter in filters:
if show_info:
print('stabilisation de la position pos %i' % pos)
- print('stabilisation du filtre %i avec le déplacement filtre %i' % (img_filter, drift_filter))
- for ti, t in enumerate(range(1, len(imgs_stack))):
- imgs_stack[t, pos, 0,:,:,img_filter] = ndi.shift(imgs_stack[t,pos,0,:,:,img_filter], dxdy[ti])
-
+ print('stabilisation du filtre %i' % img_filter)
+
+ imgref = imgs_stack[0, pos, 0, 5:-5, 5:-5, img_filter]
+
+ if drift_algo == 'sift':
+ print('Use SIFT to align stack')
+ try:
+ dxdyt = Parallel(n_jobs=-1, backend='multiprocessing')(delayed(register_task_sift)(imgs_stack[t,pos,0, 5:-5, 5:-5, img_filter],
+ imgs_stack[t-1,pos,0, 5:-5, 5:-5, img_filter], remove_cosmics=remove_cosmics) for t in range(1, len(imgs_stack)))
+ dxdyt = np.cumsum(dxdyt, axis=0).tolist()
+
+ except:
+ print('Error in SIFT, failback to crosscorrelation registration')
+ drift_algo = 'cor'
+
+ if drift_algo == 'cor':
+ print('Use crosscorrelation algorithm to align stack')
+ if remove_cosmics:
+ el = disk(2)
+ imgref = opening(imgref, el)
+
+ if gaussian_size > 0:
+ imgref = gaussian(imgref, gaussian_size)
+
+ # Non parrallel version
+ # dxdyt = [register_task(imgs_stack[t,pos,0,:,:,img_filter], imgref, up_factor=up_factor) for t in range(1, len(imgs_stack))]
+ # Parralel version
+ if remove_cosmics:
+ dxdyt = Parallel(n_jobs=-1, backend='multiprocessing')(delayed(register_task)(opening(imgs_stack[t,pos,0, 5:-5, 5:-5, img_filter], el),
+ imgref, up_factor=up_factor) for t in range(1, len(imgs_stack)))
+ else:
+ dxdyt = Parallel(n_jobs=-1, backend='multiprocessing')(delayed(register_task)(imgs_stack[t,pos,0, 5:-5, 5:-5, img_filter],
+ imgref, up_factor=up_factor) for t in range(1, len(imgs_stack)))
+
+ dxdy += dxdyt
+
+ if drift_filter is None:
+ for i, t in enumerate(range(1, len(imgs_stack))):
+ imgs_stack[t, pos, 0,:,:,img_filter] = ndi.shift(imgs_stack[t,pos,0,:,:,img_filter], dxdyt[i])
+
+ if drift_filter is not None:
+ for img_filter in range(imgs_stack.shape[5]):
+ if show_info:
+ print('stabilisation de la position pos %i' % pos)
+ print('stabilisation du filtre %i avec le déplacement filtre %i' % (img_filter, drift_filter))
+ for ti, t in enumerate(range(1, len(imgs_stack))):
+ imgs_stack[t, pos, 0,:,:,img_filter] = ndi.shift(imgs_stack[t,pos,0,:,:,img_filter], dxdy[ti])
+
# A second drift correction between filters, only on the first time
if register_filter:
+ print('Register filters between two filters !')
filter_ref = ref_filter
filter_move = move_filter
for pos in range(imgs_stack.shape[1]):
@@ -369,8 +450,8 @@ def register_images(imgs_stack, gaussian_size = 0,
if show_info:
print('Error: %0.2e' % (error))
- for t in range(0, len(imgs_stack)):
- imgs_stack[t,pos,0,:,:,filter_move] = ndi.shift(imgs_stack[t,pos,0,:,:,filter_move], dxdyt)
+ for t in range(0, len(imgs_stack)):
+ imgs_stack[t,pos,0,:,:,filter_move] = ndi.shift(imgs_stack[t,pos,0,:,:,filter_move], dxdyt)
return imgs_stack
@@ -449,7 +530,11 @@ def bacteria_bg(bacteria_contour, image):
bg_values = np.empty(len(bcontours))
# Estimation of camera bg to fix a lower bound
- camera_bg = np.percentile(np.ma.masked_equal(image, 0), (0.2,99.8))[0]
+ maimage = np.ma.masked_equal(image, 0)
+ if len(maimage[~maimage.mask]) > 0:
+ camera_bg = np.percentile(maimage[~maimage.mask], (0.2,99.8))[0]
+ else:
+ camera_bg = 0
for i in range(len(bcontours)):
@@ -623,7 +708,7 @@ def segment_bactos(image, image_band=1, denoise=True,
-def segment_bactos_stardist(image, image_band=1, model=None, prob_thresh):
+def segment_bactos_stardist(image, image_band=1, model=None, prob_thresh=None):
"""
Use StarDist trained on TELEMOS bactos to segment bacteria on fluorescence images.
@@ -644,7 +729,7 @@ def segment_bactos_stardist(image, image_band=1, model=None, prob_thresh):
"""
- # Select image
+ # Select image
img = image[:,:,image_band]
# Normalise image
@@ -788,9 +873,9 @@ def compute_bactos_properties(labels, images, position=0, pix_per_micron=1,
bactos_bbox = [r.bbox for r in regprops]
bactos_labels = [r.label for r in regprops]
+ el = disk(2)
if fluo_bg == 'auto':
process_fluo = True
- el = disk(2)
bacteria_contours = enlarged_bacteria_contour(labels)
else:
process_fluo = False
@@ -798,13 +883,16 @@ def compute_bactos_properties(labels, images, position=0, pix_per_micron=1,
bactos_mean_intensity = np.empty((len(images),len(np.unique(labels))-1, len(channels)))
for t in range(len(images)):
for chan in channels:
- regs = regionprops(labels, intensity_image=images[t, position, 0, :, :, chan])
+
+ # opening filter to remove cosmics!
+ imgt = opening(images[t, position, 0, :, :, chan], el)
+
+
+ regs = regionprops(labels, intensity_image=imgt)
for i, r in enumerate(regs):
bactos_mean_intensity[t,r.label-1, channels.index(chan)] = r.mean_intensity
if process_fluo:
- # opening filter to remove cosmics!
- imgt = opening(images[t, position, 0, :, :, chan], el)
fluo_bg = bacteria_bg(bacteria_contours, imgt)
bactos_mean_intensity[t, :, channels.index(chan)] = bactos_mean_intensity[t, :, channels.index(chan)] - fluo_bg
@@ -817,7 +905,8 @@ def extract_bactos(imgs_stack, mask_band=1, denoise=True,
pix_per_micro=1, use_stardist=False,
stardist_modelname = 'stardistBactosTELEMOS',
stardist_modeldir = '../stardist/models/',
- background=700, channels=[0, 1], prob_thresh=None):
+ background=700, channels=[0, 1], prob_thresh=None,
+ time_to_extract_bacteria=0):
"""
Make mask on selected band for the time=0 and extract intensity
@@ -856,6 +945,9 @@ def extract_bactos(imgs_stack, mask_band=1, denoise=True,
- prob_thresh, None or a float value btwn 0 1:
Set the probability threshold for stadist predict (use None to use the best one from the training)
+
+ - time_to_extract_bacteria, int:
+ The image to use in the time serie to extract the bacteria, the default is the first one so 0.
"""
data = {}
@@ -871,10 +963,10 @@ def extract_bactos(imgs_stack, mask_band=1, denoise=True,
tmpd = {}
if use_stardist:
- labels = segment_bactos_stardist(imgs_stack[0,p,0], image_band=mask_band,
+ labels = segment_bactos_stardist(imgs_stack[time_to_extract_bacteria,p,0], image_band=mask_band,
model=stardist_model, prob_thresh=prob_thresh)
else:
- labels = segment_bactos(imgs_stack[0,p,0], image_band=mask_band, denoise=denoise,
+ labels = segment_bactos(imgs_stack[time_to_extract_bacteria,p,0], image_band=mask_band, denoise=denoise,
min_size=bact_size[0], max_size=bact_size[1], min_eccentricity=bact_eccent[0],
max_eccentricity=bact_eccent[1], pix_per_micron=pix_per_micro)
@@ -1028,7 +1120,10 @@ def save_to_h5(output_file, imgs_stack, bactos_data, img_metadata, gp_name,
def run_processing(data_list, nom_sortie, MM_version=2, background='auto',
scale_correction=False, drug_filter=0, ctrl_filter=1,
- stardist_path = None, bact_size=(0.5, 15), prob_thresh=None):
+ stardist_path = None, bact_size=(0.5, 15), prob_thresh=None,
+ time_to_extract_bacteria=0, build_bacteria_figures=True,
+ bacteria_extract_mask='TRP', drift_algo='cor', register_filter=True,
+ register_stack=True, register_mask=True, drift_filter=0):
"""
Loop over the given folder name to extract data and export them inside one hdf file (containing all data from each folder)
@@ -1058,6 +1153,34 @@ def run_processing(data_list, nom_sortie, MM_version=2, background='auto',
- prob_thresh, None or a float value bitween (0, 1)
The probability threshold used by stardist predict (if None, use the best one from the training dataset)
+ - time_to_extract_bacteria, int:
+ The image to use in the time serie to extract the bacteria, the default is the first one so 0.
+
+ - build_bacteria_figures, bool:
+ Add a figure for each bacteria to the hdf5 file in an svg format. This figure contains a summary for each bacteria.
+
+ - bacteria_extract_mask, str in ['TRP' or 'DRUG']:
+ Give the mask to extract the mask of bacteria. The default 'TRP' is on the control filter
+ (the one of the tryptophan). If 'DRUG' is given the position of drug_filter is used
+
+ - drift_algo, str in ['cor', 'sift']:
+ Define the algorithm to compute the drift between images.
+ - 'cor': use crosscorrelation (the default)
+ - 'sift': use SIFT point finding + RANSAC to find transform matrix
+
+ - register_filter, bool:
+ Register the drug_filter on the ctrl_filter (default True)
+
+ - register_stack, bool:
+ Register the image stack using *drift_algo*, (default True). If false the
+ raw stack is used.
+
+ - register_mask, bool:
+ Register the two filter based on bacteria mask (default True)
+
+ - drift_filter, bool:
+ The filter on which to register the images in time
+
Example:
--------
@@ -1069,7 +1192,9 @@ def run_processing(data_list, nom_sortie, MM_version=2, background='auto',
run_processing(expes, out_file)
"""
- for c in data_list:
+ #for i in track(range(len(data_list)), description="Processing data..."):
+ for i in range(len(data_list)):
+ c = data_list[i]
datap = c
last_dir = os.path.normpath(c).split(os.path.sep)[-1]
print('Process %s, use name %s' % (c, last_dir))
@@ -1109,32 +1234,49 @@ def run_processing(data_list, nom_sortie, MM_version=2, background='auto',
print('stabilisation')
imgs = register_images(imgs, gaussian_size=0, move_filter=drug_filter,
- ref_filter=ctrl_filter, register_filter=True,
- show_info=False, up_factor=10, drift_filter=0,
- scale_correction=scale_correction)
+ ref_filter=ctrl_filter, register_filter=register_filter,
+ register_stack=register_stack, show_info=False, up_factor=10, drift_filter=drift_filter,
+ scale_correction=scale_correction,
+ drift_algo=drift_algo)
# Extract bacteria
data = {}
+ mask_band = ctrl_filter
+ filter_ref = ctrl_filter
+ filter_target = drug_filter
+ if bacteria_extract_mask == 'DRUG':
+ mask_band = drug_filter
+ filter_ref = drug_filter
+ filter_target = ctrl_filter
+
bdata = extract_bactos(imgs, bact_size=bact_size, pix_per_micro=pix_per_micro,
use_stardist=True, stardist_modeldir=stardist_path,
- mask_band=ctrl_filter,
+ mask_band=mask_band,
channels = sorted([drug_filter, ctrl_filter]),
- prob_thresh=prob_thresh)
+ prob_thresh=prob_thresh,
+ time_to_extract_bacteria=time_to_extract_bacteria,
+ background=background)
# Estimate position between two filters using bacteria pattern matching
- for pos in range(imgs.shape[1]):
- print('Resgister filter1 and filter0 for pos %i' % pos)
- imgs = register_filter_from_bactos(bdata['POS%i'%pos]['bmask'], imgs, position=pos,
- filter_ref=ctrl_filter, filter_target=drug_filter)
-
- # Update intensity data on re-registred images
- bdata = update_bactos_data(bdata, imgs, pix_per_micro, background=background,
- channels = sorted([drug_filter, ctrl_filter]))
+ if register_mask:
+ if ctrl_filter != drug_filter:
+ for pos in range(imgs.shape[1]):
+ print('Resgister filter1 and filter0 for pos %i' % pos)
+ imgs = register_filter_from_bactos(bdata['POS%i'%pos]['bmask'], imgs, position=pos,
+ filter_ref=filter_ref, filter_target=filter_target)
+
+ # Update intensity data on re-registred images
+ bdata = update_bactos_data(bdata, imgs, pix_per_micro, background=background,
+ channels = sorted([drug_filter, ctrl_filter]))
if len(bdata['POS0']['labels']) + len(bdata['POS1']['labels']) > 0:
# Sauvegarde
save_to_h5(nom_sortie, imgs, bdata, metas, last_dir, pix_per_micro,
drug_filter=drug_filter, ctrl_filter=ctrl_filter)
+
+ if build_bacteria_figures:
+ for pos in range(imgs.shape[1]):
+ save_bacterias_svg_figures(nom_sortie, last_dir, pos)
else:
print('No bacteria found, no need to save the data')
@@ -1239,3 +1381,251 @@ def correct_scaling(img_stack, ref_image, moving_image):
img_stack = register_stack_for_scaling(img_stack, moving_image, scaling)
return img_stack
+
+
+def get_bacteria_bbox(mask, bacteria_id, region_props=None, area_expend=0.5):
+ """
+ Return the coordinate of the bounding box of bacteria
+
+ Paremeters:
+ -----------
+
+ - mask, 2D array:
+ The mask of bacteria
+
+ - bacteria_id, int:
+ The bacteria that we want on the mask
+
+ - region_props, None or scikit image regionprops object:
+ The region props object for the mask could be passed to the function.
+ If its None, the region_props is evaluated using scikit image
+
+ return xmin, xmax, ymin, ymax
+ """
+
+ if region_props is None:
+ region_props = regionprops(mask)
+
+ r = region_props[bacteria_id]
+
+ # Get crops around a bacteria
+ bxmin, bymin, bxmax, bymax = r.bbox
+ bheight = bxmax - bxmin
+ bwidth = bymax - bymin
+ bfcy, bfcx = r.centroid
+ offset = area_expend
+ maxsize = max((bwidth, bheight))
+ new_size = maxsize + maxsize*offset
+
+ cropxmin = int(bfcx - new_size/2)
+ cropxmax = int(bfcx + new_size/2)
+ cropymin = int(bfcy - new_size/2)
+ cropymax = int(bfcy + new_size/2)
+
+ if cropxmin < 0:
+ cropxmin = 0
+ if cropymin < 0:
+ cropymin = 0
+
+ if cropxmax > mask.shape[0]:
+ cropxmax = mask.shape[0]
+ if cropymax > mask.shape[1]:
+ cropymax = mask.shape[1]
+
+ return cropxmin, cropxmax, cropymin, cropymax
+
+
+def bacteria_image_wall(imgs, pos, crop, channels=None, percentil=(1, 99)):
+ """
+ Create a unique image with all channels and all time of the bacteria
+
+ Parameters:
+ -----------
+
+ - imgs, np array:
+ array of images in the format [time, pos, z, y, x, channel]
+
+ - pos, int:
+ the position to use on the imgs stack
+
+ - crop, list of int of size 4:
+ the spatial limit of the stack defined as [xmin, xmax, ymin, ymax]
+
+ -channels, None or list of channel position:
+ Position of channel to stack vertically in the image_wall. If None all channel will be used.
+
+ - percentil, tuple of size 2:
+ the min percentil and max percentil to crop the intensities of imgs
+ for autoscaling.
+
+ return array
+ """
+
+ cropxmin, cropxmax, cropymin, cropymax = crop
+
+ if channels is None:
+ channels = list(range(imgs.shape[5]))
+
+ imghstack = np.hstack(imgs[:, pos, 0, cropymin:cropymax, cropxmin:cropxmax, channels[0]])
+ ps, pe = np.percentile(imghstack, percentil)
+ imgwall = exposure.rescale_intensity(imghstack, in_range=(ps, pe))
+
+ if len(channels) > 1:
+ for ch in channels[1:]:
+ imghstack = np.hstack(imgs[:,pos,0, cropymin:cropymax, cropxmin:cropxmax, ch])
+ ps, pe = np.percentile(imghstack, percentil)
+ imgwall = np.vstack((imgwall, exposure.rescale_intensity(imghstack, in_range=(ps, pe))))
+
+ return imgwall
+
+
+def bacteria_summary_graph(bacteria_ids, h5data, experiment, position,
+ bacteria_offset=0.5, export_as_svg=True):
+ """
+ Build a graphic thats shows pictures time series of a given bacteria
+ and the evolution of it's fluorescence.
+
+ Parameters:
+ -----------
+
+ - bacteria_ids, (int or float) or 'all' or a list of int or float:
+ The bacteria id of list of bacteria ids to use. If 'all' take the unique
+ value of the mask to get all bacterias.
+
+ - h5data, str:
+ The path to the data in hdf format
+
+ - experiment, str:
+ The name of the experiment. Must be contained in the h5data
+
+ - position, int:
+ The position, if multi stage position has been used on the microscope, to use
+
+ - bacteria_offset, float:
+ The size offset (in percent) to add around the bacteria bounding box
+
+ - export_as_svg, bool:
+ Export the figure as svg format if True (default). If False return a list of
+ matplotlib figure objects.
+
+ return a list of figures
+ """
+
+
+ if isinstance(bacteria_ids, (int, float)):
+ bacteria_ids = [bacteria_ids]
+
+ h5data = Path(h5data)
+
+ # Load the data
+ pos = position
+ exp = experiment
+ imgs = load_image(h5data, exp)
+ masks = load_mask(h5data, exp, pos)
+ bactosct = bacteria_contour(masks)
+ Ib_chan0 = load_intensity_profile(h5data, exp, 0, pos)
+ Ib_chan1 = load_intensity_profile(h5data, exp, 1, pos)
+
+ if isinstance(bacteria_ids, str) and bacteria_ids == 'all':
+ bacteria_ids = np.unique(masks)[1:]-1
+
+ # Compute the properties of bacteria masks
+ rprops = regionprops(masks)
+
+ # This list will contain all figures (one for each bacteria)
+ figures = []
+
+ for bid in bacteria_ids:
+ r = rprops[bid]
+
+ # Get crops around a bacteria
+ cropxmin, cropxmax, cropymin, cropymax = get_bacteria_bbox(masks, bid, rprops, bacteria_offset)
+ # Compute the new width and height for this bacteria
+ nwidth = cropxmax - cropxmin
+ nheight = cropymax - cropymin
+
+ # Get the centroid of bacteria
+ bfcy, bfcx = r.centroid
+
+ # Start to build the figure
+ fig = plt.figure(figsize=(13, 1.5))
+ plt.title(f'Exp: {exp}, Pos: {pos}, bacteria-{bid+1} (in file {h5data.name})')
+ G = plt.GridSpec(1, 9)
+
+ # The main image to locate the bacteria on the FOV
+ plt.subplot(G[0, 0])
+ p2, p98 = np.percentile(imgs[0,pos,0, ::2, ::2, 0], (2, 98))
+ plt.imshow(exposure.rescale_intensity(imgs[0,pos,0, ::2, ::2, 0], in_range=(p2, p98)), 'gray')
+ plt.plot(bfcx/2, bfcy/2, 'ro', fillstyle='none')
+ plt.axis('off')
+ plt.text(-8, imgs.shape[4]/4, f'Bacteria-{bid+1}', rotation=90, ha='right', va='center')
+ plt.text(imgs.shape[4]/4, 0, f'POS - {pos}', ha='center', va='bottom')
+
+ # Create the timewall of all images of this bacteria through time and filters
+ plt.subplot(G[0, 1:4])
+ imgwall = bacteria_image_wall(imgs, pos, (cropxmin, cropxmax, cropymin, cropymax))
+ plt.imshow(imgwall, 'gray')
+
+ # Add channel names
+ for ch in range(imgs.shape[5]):
+ plt.text(-5, nheight/2+ch*nheight, f'C-{ch}', ha='right', va='center')
+
+ # Plot the contour of bacteria on first row of the timewall and for all time steps
+ for j in range(len(imgs)):
+ plt.plot((bactosct[bid][1] + j*nwidth)-cropxmin, bactosct[bid][0]-cropymin, 'C3--')
+
+ plt.axis('off')
+
+ # Plot the mean intensity of each filters inside the segmented bacteria
+ plt.subplot(G[0, 4:])
+ plt.plot(Ib_chan1[:, bid], 'C3o-')
+ ax = plt.gca()
+ ax.spines["right"].set_visible(False)
+ ax.spines["top"].set_visible(False)
+ ax.set_ylabel('I-C-0', color='C3')
+ ax.tick_params(axis='y', labelcolor='C3')
+
+ ax2 = ax.twinx()
+ ax2.plot(Ib_chan0[:, bid], 'C0o-')
+ ax2.spines["left"].set_visible(False)
+ ax2.spines["top"].set_visible(False)
+ ax2.tick_params(axis='y', labelcolor='C0')
+ ax2.set_ylabel('I-C-1', color='C0')
+
+
+ plt.tight_layout(pad=0.9)
+
+ if export_as_svg:
+ fsvg = io.BytesIO()
+ plt.savefig(fsvg, format='svg')
+ figures += [fsvg.getvalue().strip().replace(b'\n', b'')]
+ plt.close(fig)
+ del fig
+ else:
+ figures += [fig]
+
+ return figures
+
+
+def save_bacterias_svg_figures(h5file, exp, pos):
+ """
+ Save figures in svg format for each bacteria, the figure shows a closed view of the
+ bacteria it's location and the signal inside the mask with time.
+ """
+
+ # create all the figures
+ print('Create svg figures in Parralel')
+ nbactos = np.unique(load_mask(h5file, exp, pos))[1:]-1
+ figs = Parallel(n_jobs=-1, backend='multiprocessing')(delayed(bacteria_summary_graph)(int(b), h5file, exp, pos) for b in nbactos)
+ figs = [f[0] for f in figs]
+
+ # Not in parralel we can use the following function
+ # figs = bacteria_summary_graph('all', h5file, exp, pos)
+
+ print(f'Save svg figures to {exp}/{pos}/bacteria_svg_figures in {h5file}')
+ with h5py.File(h5file, 'a') as f:
+ pos = 'POS%i' % pos
+ if 'bacteria_svg_figures' in f[exp][pos]:
+ del f[exp][pos]['bacteria_svg_figures']
+
+ f[exp][pos].create_dataset('bacteria_svg_figures', data=figs, compression="gzip")
diff --git a/src/bactos_viewver.py b/src/bactos_viewver.py
index 8c7b67f03894e900f5b0530e721bcc26224214a5..faa693db9f33ab07c63fefa66d5de02e1a88ab37 100755
--- a/src/bactos_viewver.py
+++ b/src/bactos_viewver.py
@@ -3,7 +3,7 @@ Use ipywidgets and ipympl to create a graphical user interface inside jupyter
notebook to check bacteria drug uptake on DISCO UV fluorescence image stack.
"""
-import ipywidgets as widgets
+from ipywidgets import widgets, GridspecLayout, Layout
import numpy as np
import matplotlib.pyplot as plt
from matplotlib.collections import PatchCollection
@@ -18,13 +18,14 @@ import os
import shutil
import logging
from matplotlib.backend_bases import NavigationToolbar2, Event
+from base64 import b64encode
OLDHOME = NavigationToolbar2.home
def new_home(self, *args, **kwargs):
# register a new event
OLDHOME(self, *args, **kwargs)
- event = Event('home_event', self)
- self.canvas.callbacks.process('home_event', event)
+ # event = Event('home_event', self)
+ # self.canvas.callbacks.process('home_event', event)
# redefine home command
NavigationToolbar2.home = new_home
@@ -73,7 +74,7 @@ def get_all_pos_fluo(selected_file, selected_exp, limit_to_selected_bactos=True,
I0 = I0[:, selected_bactos]
I1 = I1[:, selected_bactos]
- # LE TRP est sur le channel 1 et le filtre 2 sur le channel 0
+ # LE TRP est sur le channel 1 et le filtre 2 sur le channel 0
IF1 += [I1]
IF2 += [I0]
@@ -459,6 +460,8 @@ class BactosViewver():
self.Wimgout = widgets.Output()
self.WplotAll = widgets.Output()
self.Woutbactos = widgets.Output(layout=widgets.Layout(width="330px"))
+ self.BacteriaPlots = widgets.Output(layout=Layout(min_height="600px"))
+ self.bacteria_grid = None
self.BactosAreaOutput = widgets.Output()
self.I1output = widgets.Output()
self.I2output = widgets.Output()
@@ -535,9 +538,9 @@ class BactosViewver():
# Create the tab pannels
tabs = widgets.Tab() # layout=widgets.Layout(width="75%")
- tabs.children = [wimgs, self.AllPosPlotOutput, self.AllExpPlotOutput, self.BactosAreaOutput,
+ tabs.children = [wimgs, self.AllPosPlotOutput, self.AllExpPlotOutput, self.BacteriaPlots, self.BactosAreaOutput,
self.I1output, self.I2output, self.I1overI2output, self.DrugIoutput]
- tabs_titles = ('Plot: data explorer', 'Plot: Idrug for all pos', 'Plot: all exp', 'Data: Bactos area',
+ tabs_titles = ('Plot: data explorer', 'Plot: Idrug for all pos', 'Plot: all exp', 'Plot: bacteria summary','Data: Bactos area',
'Data: F2 fluo', 'Data: F1 fluo', 'Data: F2/F1 control', 'Data: Drug fluo')
for i, title in enumerate(tabs_titles):
@@ -560,6 +563,7 @@ class BactosViewver():
footer=None)
tabs.observe(self.on_tab_changed, names='selected_index')
+ self.tabs = tabs
display(self.final_lay)
@@ -646,10 +650,18 @@ class BactosViewver():
# Ajust min-max
if adjust_minmax:
- self.F2minmax.value = list(np.percentile(np.ma.masked_equal(self.selected_imgs[0, 0, 0, :, :, self.drug_filter], 0),
- (0.2, 99.9)))
- self.F1minmax.value = list(np.percentile(np.ma.masked_equal(self.selected_imgs[0, 0, 0, :, :, self.ctrl_filter], 0),
+ try:
+ self.F2minmax.value = list(np.percentile(np.ma.masked_equal(self.selected_imgs[0, 0, 0, :, :, self.drug_filter], 0),
+ (0.2, 99.9)))
+ except:
+ pass
+
+ try:
+ self.F1minmax.value = list(np.percentile(np.ma.masked_equal(self.selected_imgs[0, 0, 0, :, :, self.ctrl_filter], 0),
(0.2, 99.9)))
+ except:
+ pass
+
else:
self.plimg1.set_clim(self.F2minmax.value)
self.plimg2.set_clim(self.F1minmax.value)
@@ -665,7 +677,8 @@ class BactosViewver():
# Connect the update of profiles based on bacteria currently displayed on the image
self.figure.canvas.mpl_connect('button_release_event', self.update_bactos_inview)
- self.figure.canvas.mpl_connect('home_event', self.update_bactos_inview)
+ # TODO: MPL find new way to triger home event (it's no more available)
+ # self.figure.canvas.mpl_connect('home_event', self.update_bactos_inview)
with self.WplotAll:
plt.tight_layout()
@@ -676,7 +689,7 @@ class BactosViewver():
drawing each contour individually).
"""
- # get figure axes
+ # get figure axes
ax1, ax2, ax3, ax4, ax5, ax6 = self.figure.get_axes()
if force_redraw:
@@ -1106,7 +1119,7 @@ class BactosViewver():
self.figure_all_exp = plt.figure('Plot all experiments', figsize=(11,5))
plt.ion()
- # Calcul du nombre d'expérience avec un contrôle
+ # Calcul du nombre d'expérience avec un contrôle
# TODO !!!
ax1 = plt.subplot(111)
@@ -1259,6 +1272,16 @@ class BactosViewver():
sel = len([i for i, val in enumerate(self.selected_bactos) if val])
self.accordion.set_title(1, 'Select bactos (%i/%i)' % (sel, tot) )
+ # Update the bacteria plot summary border if needed
+ if self.bacteria_grid is not None:
+ cb = self.bactos_colors[id_bactos] * 255
+ if new_state:
+ bstyle = f"2px solid rgb({cb[0]}, {cb[1]}, {cb[2]})"
+ else:
+ bstyle = "2px dashed gray"
+
+ self.bacteria_grid[id_bactos, 0].layout.border = bstyle
+
#Update tables
self.update_area_table()
self.update_intensity_tables()
@@ -1323,7 +1346,7 @@ class BactosViewver():
self.H = len(self.selected_imgs[0,0,0,0,:,0])
self.bactos_dist_slider.max = min(self.W, self.H)/2 - 10
- # Need to update the extent
+ # Need to update the extent
self.EXTENT = (0, self.H, self.W, 0)
self.plimg1.set_extent(self.EXTENT)
self.plimg2.set_extent(self.EXTENT)
@@ -1350,7 +1373,7 @@ class BactosViewver():
self.ListPos.options = self.positions
self.ListPos.value = self.selected_pos
- # Load the control experiment if it as been given
+ # Load the control experiment if it as been given
ctrl_exp = load_control_exp(self.selected_file, selected)
# print(ctrl_exp)
if ctrl_exp is not None:
@@ -1367,7 +1390,7 @@ class BactosViewver():
# update bactos contours
self.extract_bactos_contour_coordinate()
- # print(selected_file, selected, selected_pos)
+ # print(selected_file, selected, selected_pos)
self.i1 = load_intensity_profile(self.selected_file, selected, 0, self.selected_pos)
self.i2 = load_intensity_profile(self.selected_file, selected, 1, self.selected_pos)
@@ -1380,32 +1403,34 @@ class BactosViewver():
self.update_bactos_area()
self.update_bactos_area_minmax()
- try:
- self.display_stack(self.selected_time, self.selected_pos, 0, 0, True)
- except Exception as e:
- print('Error to display images')
- print(e)
+ if self.tabs.selected_index == 0:
+ try:
+ self.display_stack(self.selected_time, self.selected_pos, 0, 0, True)
+ except Exception as e:
+ print('Error to display images')
+ print(e)
- try:
- self.display_bactos_contour_fast(force_redraw=True)
- except Exception as e:
- print('Error to display bactos contour')
- print(e)
-
- if sum(self.selected_bactos):
try:
- self.plot_profiles()
+ self.display_bactos_contour_fast(force_redraw=True)
except Exception as e:
- print('Error in plot fluo profiles')
+ print('Error to display bactos contour')
print(e)
+
+ if sum(self.selected_bactos):
+ try:
+ self.plot_profiles()
+ except Exception as e:
+ print('Error in plot fluo profiles')
+ print(e)
- with self.AllPosPlotOutput:
- try:
- self.update_all_pos_plot()
- except Exception as e:
- print('Error')
- print(e)
+ if self.tabs.selected_index == 1:
+ with self.AllPosPlotOutput:
+ try:
+ self.update_all_pos_plot()
+ except Exception as e:
+ print('Error')
+ print(e)
#Update tables
@@ -1413,8 +1438,12 @@ class BactosViewver():
self.update_intensity_tables()
self.update_control_ratio_table()
- self.Wimgout.clear_output()
+ if self.tabs.selected_index == 3:
+ self.plot_bacteria_svg_figures()
+ # Message container need to bee cleaned
+ self.Wimgout.clear_output()
+
def on_pos_changed(self, evt):
"""
Action triggered when the user select a position of the current experiement.
@@ -1443,10 +1472,15 @@ class BactosViewver():
self.update_bactos_area_minmax()
# Display new data
- self.display_stack(self.selected_time, self.selected_pos, 0, 0, True)
- self.display_bactos_contour_fast(force_redraw=True)
- self.plot_profiles()
+ if self.tabs.selected_index == 0:
+ self.display_stack(self.selected_time, self.selected_pos, 0, 0, True)
+ self.display_bactos_contour_fast(force_redraw=True)
+ self.plot_profiles()
+ # Update single bacteria figure
+ if self.tabs.selected_index == 3:
+ self.plot_bacteria_svg_figures()
+
#Update tables
self.update_area_table()
self.update_intensity_tables()
@@ -1458,7 +1492,9 @@ class BactosViewver():
new_ctrl_exp = evt['new']
save_control_exp(self.selected_file, self.ListExp.value, new_ctrl_exp)
- self.plot_profiles()
+
+ if self.tabs.selected_index == 0:
+ self.plot_profiles()
# update the table
self.update_control_ratio_table()
@@ -1594,6 +1630,37 @@ class BactosViewver():
for a in (ax4, ax5, ax6):
ax.set_xlim(0, self.selected_imgs.shape[0])
+ def plot_bacteria_svg_figures(self):
+ """
+ Update the plot of each bacteria summary
+ """
+ with self.BacteriaPlots:
+ self.BacteriaPlots.clear_output()
+ try:
+ figs = load_bacterias_svg_figures(self.selected_file, self.ListExp.value, self.selected_pos)
+ except Exception as e:
+ print('error', e)
+ figs = None
+ self.bacteria_grid = None
+
+ if figs is not None:
+ # figs = bacteria_summary_graph('all', fpath, exp, pos)
+ self.bacteria_grid = GridspecLayout(len(figs), 1, layout=Layout(height="800px"))
+ for i, fig in enumerate(figs):
+ dd = b64encode(fig)
+ url = f'data:image/svg+xml;base64,{dd.decode("utf8")}'
+ img_url = f'<img src="{url}" style="width: 100%; height: 100%;" >'
+ self.bacteria_grid[i, 0] = widgets.HTML(value=img_url)
+ cb = self.bactos_colors[i] * 255
+ if self.selected_bactos[i]:
+ bstyle = f"2px solid rgb({cb[0]}, {cb[1]}, {cb[2]})"
+ else:
+ bstyle = "2px dashed gray"
+
+ self.bacteria_grid[i, 0].layout.border = bstyle
+
+ display(self.bacteria_grid)
+
def on_time_changed(self, evt):
"""
update the time varible
@@ -1655,6 +1722,9 @@ class BactosViewver():
self.AllExpPlotOutput.clear_output(wait=True)
self.update_all_exp_plot()
plt.show(self.figure_all_exp)
+
+ if tabid == 3:
+ self.plot_bacteria_svg_figures()
def on_export(self, btn):
"""
@@ -1672,7 +1742,7 @@ class BactosViewver():
# Output file name (should be the same as the imput source)
fout = os.path.abspath(self.source_file)
fout = os.path.splitext(fout)[0] + '.xlsx'
-
+ prevsname = []
with p.ExcelWriter(fout, engine='xlsxwriter') as writer:
for expe in self.ListExp.options:
ctrl_exp = load_control_exp(self.selected_file, expe)
diff --git a/src/lib_bactos_hdf.py b/src/lib_bactos_hdf.py
index 22caca9044a430d487f84ccb1d9447f51e5a0f00..1b40c844c5929bf0f09825baa523858be6982111 100755
--- a/src/lib_bactos_hdf.py
+++ b/src/lib_bactos_hdf.py
@@ -47,7 +47,7 @@ def load_image(data_in, expe_name):
with h5py.File(data_in, 'r') as f:
output = f[expe_name]['images'][:]
-
+
return output
@@ -74,7 +74,7 @@ def load_mask(data_in, expe_name, pos):
with h5py.File(data_in, 'r') as f:
pos = 'POS%i' % pos
output = f[expe_name][pos]['bacteria_mask'][:]
-
+
return output
def get_stack_shape(data_in, expe_name):
@@ -94,7 +94,7 @@ def get_stack_shape(data_in, expe_name):
"""
with h5py.File(data_in, 'r') as f:
output = f[expe_name]['images'].shape
-
+
return output
def load_intensity_profile(data_in, expe_name, channel, pos):
@@ -123,7 +123,7 @@ def load_intensity_profile(data_in, expe_name, channel, pos):
with h5py.File(data_in, 'r') as f:
pos = 'POS%i' % pos
output = f[expe_name][pos]['ch%i_bacteria_intensities' % channel][:]
-
+
return output
@@ -149,8 +149,8 @@ def load_area(data_in, expe_name, pos):
with h5py.File(data_in, 'r') as f:
pos = 'POS%i' % pos
output = f[expe_name][pos]['bacteria_area'][:]
-
- return output
+
+ return output
def save_selected_bactos(data_in, expe_name, pos, selected_bactos):
@@ -205,7 +205,7 @@ def save_drug_signal(data_in, expe_name, pos, drug_data):
pos = 'POS%i' % pos
if 'drug_signal' in f[expe_name][pos]:
del f[expe_name][pos]['drug_signal']
-
+
f[expe_name][pos].create_dataset('drug_signal', data=drug_data)
@@ -228,16 +228,16 @@ def load_drug_signal(data_in, expe_name, pos):
"""
output = None
-
+
with h5py.File(data_in, 'r') as f:
pos = 'POS%i' % pos
-
+
if 'drug_signal' in f[expe_name][pos]:
output = f[expe_name][pos]['drug_signal'][:]
-
+
return output
-
-
+
+
def load_selected_bactos(data_in, expe_name, pos):
"""
Get the list of user selected bacteria save in h5file for a given experiment
@@ -264,7 +264,7 @@ def load_selected_bactos(data_in, expe_name, pos):
output = f[expe_name][pos]['selected_bactos'][:]
else:
output = None
-
+
return output
@@ -339,18 +339,17 @@ def load_filter_positions(data_in, expe_name):
print('Use pos 0 for drug and pos 1 for controle')
return ctrl_filter, drug_filter
-
-
+
def load_bacterias_svg_figures(h5file, exp, pos):
"""
Load svg figures of bacterias
"""
-
+
with h5py.File(h5file, 'r') as f:
pos = 'POS%i' % pos
if 'bacteria_svg_figures' in f[exp][pos]:
output = f[exp][pos]['bacteria_svg_figures'][:]
else:
output = None
-
+
return output