}
detected_topomesh = vertex_topomesh(seed_positions)
# -- Detect L1 using 'nuclei_layer':
oriented_seeds = {
k: np.array([1., 1., -1.]) * v
for k, v in seed_positions.items()
}
cell_layer = img_graph.vertex_property('L1')
# -- Update the topomesh with the 'layer' property:
detected_topomesh.update_wisp_property(
'layer', 0, cell_layer)
# -- Save the detected topomesh:
ppty2ply = dict([(0, ['layer']), (1, []), (2, []),
(3, [])])
save_ply_property_topomesh(detected_topomesh,
topomesh_file,
properties_to_save=ppty2ply,
color_faces=False)
# - Create a 'detected_topomesh' out of L1 cells only:
L1_detected_topomesh = filter_topomesh_vertices(
detected_topomesh, "L1")
# world.add(L1_detected_topomesh, L1_detected_seed"+suffix)
# world[L1_detected_seed"+suffix]["property_name_0"] = 'layer'
# world["L1_detected_seed{}_vertices".format(suffix)]["polydata_colormap"] = load_colormaps()['Reds']
# - Performs evaluation of detected seeds agains expert:
print "\n# Evaluation: comparison to EXPERT seeds..."
evaluation = evaluate_positions_detection(
detected_topomesh,
expert_topomesh,
max_distance=np.linalg.norm(size * vxs))
filename] + "_nuclei_detection_topomesh{}.ply".format(suffix)
if not os.path.exists(topomesh_file):
# topomesh, surface_topomesh = nuclei_image_topomesh(image_dict,threshold=1000,reference_name=reference_name,microscope_orientation=microscope_orientation,signal_names=signal_names,compute_ratios=compute_ratios,subsampling=4,return_surface=True)
topomesh = nuclei_image_topomesh(
image_dict,
threshold=1000,
reference_name=reference_name,
microscope_orientation=microscope_orientation,
signal_names=signal_names,
compute_ratios=compute_ratios,
subsampling=4,
surface_subsampling=6)
save_ply_property_topomesh(topomesh,
topomesh_file,
properties_to_save=dict([
(0, signal_names + ['layer']), (1, []),
(2, []), (3, [])
]),
color_faces=False)
df = topomesh_to_dataframe(topomesh, 0)
df.to_csv(image_dirname + "/" + nomenclature_names[filename] + "/" +
nomenclature_names[filename] +
"_signal_data{}.csv".format(suffix))
# Correction step:
for rescaling in [False, True]:
reference_img = image_dict[reference_name]
try:
vxs = reference_img.voxelsize
except:
center = np.array([size/2,size/2,size/2],float)*np.array(img.resolution)
positions = L1_topomesh.wisp_property('barycenter',0)
for p in L1_topomesh.wisps(0):
positions[p] = center + (size/3.)*np.array(img.resolution)*(positions[p]-center)/np.linalg.norm(positions[p]-center)
L1_topomesh.update_wisp_property('barycenter',0,positions)
world.add(L1_topomesh ,'dual_reconstuction')
triangular_string = ""
for t in triangular:
triangular_string += t+"_"
topomesh_filename = dirname+"/output_meshes/"+filename+"/"+filename+"_voronoi_L1_"+triangular_string+"topomesh.ply"
save_ply_property_topomesh(L1_topomesh,topomesh_filename,color_faces=True)
triangulation_topomesh = deepcopy(draco.triangulation_topomesh)
from openalea.cellcomplex.property_topomesh.property_topomesh_extraction import epidermis_topomesh, clean_topomesh
triangulation_topomesh = epidermis_topomesh(triangulation_topomesh)
triangulation_triangles = np.array([list(triangulation_topomesh.borders(2,t,2)) for t in triangulation_topomesh.wisps(2)])
not_L1_triangles = np.array(list(triangulation_topomesh.wisps(2)))[np.sum(draco.cell_layer.values(triangulation_triangles) == 1,axis=1)<2]
for t in not_L1_triangles:
triangulation_topomesh.remove_wisp(2,t)
clean_topomesh(triangulation_topomesh)
except:
pass
else:
print "Applied mask to nuclei signal...",
if iso_resampling:
print "\nPerforming 'isometric_resampling'..."
img = isometric_resampling(img)
if std_dev != 0:
print "\nPerforming 'gaussian_smoothing' with std_dev={}...".format(std_dev)
img = linear_filtering(img, 'gaussian_smoothing', std_dev=std_dev)
# - Performs nuclei detection:
detected_topomesh = nuclei_image_topomesh(dict([(nuclei_ch_name, img)]), reference_name=nuclei_ch_name, signal_names=[], compute_ratios=[], microscope_orientation=microscope_orientation, radius_range=(radius_min,radius_max), threshold=threshold, subsampling=10)
# detected_positions = detected_topomesh.wisp_property('barycenter',0)
print "Done"
save_ply_property_topomesh(detected_topomesh, topomesh_file, properties_to_save=dict([(0,[nuclei_ch_name]+['layer']),(1,[]),(2,[]),(3,[])]), color_faces=False)
# - Display detected_topomesh:
# world.add(detected_topomesh, "detected_nuclei"+suffix)
# world["detected_nuclei"+suffix]["property_name_0"] = 'layer'
# world["detected_nuclei"+suffix+"_vertices"]["polydata_colormap"] = load_colormaps()['Reds']
# - Filter L1-detected nuclei:
L1_detected_topomesh = filter_topomesh_vertices(detected_topomesh, "L1")
# - Display L1_detected_topomesh:
# world.add(L1_detected_topomesh,"L1_detected_nuclei"+suffix)
# world["L1_detected_nuclei"+suffix]["property_name_0"] = 'layer'
# world["L1_detected_nuclei"+suffix+"_vertices"]["polydata_colormap"] = load_colormaps()['Reds']
# - Evaluate nuclei detection for all cells:
print "Evaluate nuclei detection for all cells:"
#triangular = ['star','remeshed','realistic','projected','flat']
triangular = ['star', 'remeshed']
image_dual_topomesh = draco.dual_reconstruction(
reconstruction_triangulation=triangular,
adjacency_complex_degree=3,
maximal_edge_length=5.1)
#image_dual_topomesh = draco.draco_topomesh(reconstruction_triangulation = triangular)
from openalea.cellcomplex.property_topomesh.property_topomesh_optimization import property_topomesh_vertices_deformation
#property_topomesh_vertices_deformation(image_dual_topomesh,iterations=15)
world.add(image_dual_topomesh, 'dual_reconstuction')
draco_file = dirname + "/output_meshes/" + filename + "/" + filename + "_DRACO_mesh.ply"
save_ply_property_topomesh(image_dual_topomesh,
draco_file,
color_faces=True,
colormap=load_colormaps()['glasbey'])
optimized_dual_topomesh = optimize_topomesh(
image_dual_topomesh,
omega_forces={
'regularization': 0.00,
'neighborhood': 0.65,
'taubin_smoothing': 0.65,
'laplacian': 0.7,
'planarization': 0.27,
'epidermis_planarization': 0.05,
'convexity': 0.06
},
omega_regularization_max=0.01,
edge_flip=True,
spatio_temporal_properties=['barycenter'],
ignore_cells_at_stack_margins=False)
print img_graph.nb_vertices(), " Seeds detected"
seed_positions = dict([(v, img_graph.vertex_property('barycenter')[v])
for v in img_graph.vertices()])
seed_topomesh = vertex_topomesh(seed_positions)
#world.add(seed_topomesh,'seeds')
cell_layer = nuclei_layer(seed_positions, size, voxelsize)
seed_topomesh.update_wisp_property('layer', 0, cell_layer)
topomesh_file = image_dirname + filename + "_eq_seeds_topomesh.ply"
save_ply_property_topomesh(seed_topomesh,
topomesh_file,
properties_to_save=dict([
(0, ['Lti6b'] + ['layer']), (1, []),
(2, []), (3, [])
]),
color_faces=False)
df = topomesh_to_dataframe(seed_topomesh, 0)
df.to_csv(image_dirname + filename + "_signal_data.csv")
# Correction :
seed_topomesh_filename = image_dirname + filename + "_eq_seeds_topomesh.ply"
seed_topomesh = read_ply_property_topomesh(seed_topomesh_filename)
## L1 cells only
L1_cells = np.array(list(seed_topomesh.wisps(0)))[seed_topomesh.wisp_property(
'layer', 0).values() == 1]
non_L1_cells = np.array(list(seed_topomesh.wisps(0)))[
seed_topomesh.wisp_property('layer', 0).values() != 1]