5 import skimage.exposure
8 import skimage.morphology
10 def make_mask(zstack, nbins=32768):
11 thresh = skimage.filters.threshold_otsu(zstack, nbins=32768)
12 return zstack > thresh
14 def build_blob_array(stack):
15 data = [[[False for i in range(len(stack[0][0]))] for j in range(len(stack[0]))] for k in range(len(stack))]
16 for im in range(len(stack))): #was stack
17 print "processing " + str(im)
18 yx = skimage.exposure.equalize_adapthist(stack[im], clip_limit=0.00025, nbins=32768)
22 yx = (yx - zmn) / (zmx - zmn)
24 blobs_log = skimage.feature.blob_log(yx, min_sigma=1, max_sigma=3, num_sigma=20, threshold=0.025)
25 print str(len(blobs_log)) + " blobs found"
26 if len(blobs_log) > 1:
27 blobs_log[:, 2] = blobs_log[:, 2] * math.sqrt(2)
29 data[im][int(i[0])][int(i[1])] = True
32 def make_watershed(mask):
33 distance = ndi.distance_transform_edt(mask)
34 local_max = skimage.feature.peak_local_max(distance, indices=False, footprint=np.ones((31, 31)), labels=mask)
35 markers = ndi.label(local_max)[0]
36 labels = skimage.morphology.watershed(-distance, markers, mask=mask)
39 def make_gene_layers(gene_tif):
40 gene_z = gene_tif.get_tiff_array()
41 gene_s = np.arange(0, len(gene_z)+1, len(gene_z) / 6)
42 gene_r = zip(gene_s[:-1], (gene_s + 1)[1:])
43 gene_y = np.array(gene_z[gene_r[2][0]:gene_r[2][1]], dtype=np.uint16)
46 def make_nissl_watershed(nissl_tif):
47 nissl_z = nissl_tif.get_tiff_array()
48 nissl_s = np.arange(0, len(nissl_z)+1, len(nissl_z) / 6)
49 nissl_r = zip(nissl_s[:-1], (nissl_s + 1)[1:])
50 nissl_y = np.array(nissl_z[nissl_r[1][0]:nissl_r[1][1]], dtype=np.uint16)
51 nissl_watershed = [[] for k in range(len(nissl_y))]
53 for i in range(len(nissl_y)):
54 mask = make_mask(nissl_y[i])
55 watershed = make_watershed(mask)
56 nissl_watershed[i] = watershed
57 return nissl_watershed
59 def correct_labels(nissl_watershed):
60 nissl_watershed_corrected = [[[0 for i in range(len(nissl_watershed[0][0]))] for j in range(len(nissl_watershed[0]))] for k in range(len(nissl_watershed))]
61 nissl_watershed_corrected[0] = [i for i in nissl_watershed[0]]
62 nissl_watershed_corrected = np.asarray(nissl_watershed_corrected)
64 for k in range(1, len(nissl_watershed_corrected)):
65 cell_number_max = nissl_watershed_corrected[k-1].max() #changed to corrected to keep accurate count
66 cell_number_two_max = nissl_watershed[k].max()
67 and_map = np.logical_and(nissl_watershed[k-1], nissl_watershed[k])
68 and_map_pos = np.where(and_map)
69 cell_set = set(nissl_watershed[k-1][np.where(and_map)])
70 inter_layer_dict = dict()
71 for i in range(len(and_map_pos[0])):
72 inter_layer_dict[nissl_watershed[k][and_map_pos[0][i]][and_map_pos[1][i]]] = nissl_watershed[k-1][and_map_pos[0][i]][and_map_pos[1][i]]
74 inter_layer_dict[0] = 0
75 for i in range(1, cell_number_two_max + 1):
76 if not i in inter_layer_dict:
78 inter_layer_dict[i] = cell_number_max + counter
83 nissl_watershed_corrected[k][i][j] = inter_layer_dict[nissl_watershed[k][i][j]]
85 return nissl_watershed_corrected