-
Notifications
You must be signed in to change notification settings - Fork 0
/
fluowarp.py
255 lines (222 loc) · 9.17 KB
/
fluowarp.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
# -*- coding: utf-8 -*-
"""
Created on Tue Mar 4 19:53:51 2014
Phase correlation drift correction.
Used papers Cross-correlation image tracking for drift correction and
adsorbate analysis B. A. Mantooth, Z. J. Donhauser, K. F. Kelly, and P. S. Weiss
for inspiration.
@author: Monika Kauer
"""
import numpy as np
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
from scipy import ndimage
import os, sys
import scipy.misc as misc
import pump_writer
##===================================================#
# image generator in/output
## ==================================================#
def read_sequentially(params, intrvl = 1):
"""Function that reads sequentially with every call."""
filenames = os.listdir(params['directory'])
filenames = [f for f in filenames if ".%s"% params["type"] in f]
filenames = pump_writer.natural_sort(filenames)
filenames2 = filenames[params["start"]:params["end"]:intrvl]
#deal with non-divisor chunk lengths
if (params["end"]-params["start"])%intrvl != 0:
try:
filenames2.append(filenames[params["end"]])
params['nof'] = (params["end"]-params["start"])
except IndexError:
filenames2.append(filenames[-1])
params['nof'] = (len(filenames)-params["start"])
elif intrvl > 1 and (params["end"]-params["start"])%intrvl == 0:
try:
filenames2.append(filenames[params["end"]])
params['nof'] = (params["end"]-params["start"])
except IndexError:
filenames2.append(filenames[-1])
params['nof'] = (len(filenames)-params["start"])
for filename in filenames2:
yield read_img(params['directory']+filename, params)
def read_img(fname, params):
"""reads an image file as array."""
try:
image = misc.imread(fname, flatten=True)
data = np.asarray(image, dtype = np.int64)
if params['rotate']:
data = np.transpose(data)
if params['cropx']:
data = data[:,params['cropx'][0]:params['cropx'][1]]
return data
except IOError:
print fname
pass
##===================================================#
# image registration
## ==================================================#
def reg(im1, im2, params):
"""Find image-image correlation and translation vector using FFTs."""
# use hanning window. Reduces the edge effect from finite size
shape= np.array(im1.shape)
fft_im1 = np.fft.fft2(im1)
fft_im2 = np.conj(np.fft.fft2(im2))
corr = np.fft.ifft2(fft_im1*fft_im2).real
corr = ndimage.gaussian_filter(corr, .5) - ndimage.gaussian_filter(corr, 30)
t0, t1 = np.unravel_index(np.argmax(corr), shape)
if t0 > shape[0] // 2:
t0 -= shape[0]
if t1 > shape[1] // 2:
t1 -= shape[1]
return corr, [t0, 0]
def find_roi(params):
"""calculates image drift using registration via correlation."""
im = read_sequentially(params, intrvl = params["chunk"])
roi = [[0,0]]
im_new = im.next()
height, width = im_new.shape
try:
while True: #go through all image chunks from start to end
im_old = im_new
im_new = im.next()
im1 = np.where(im_old>np.median(im_old), 1,0)
im2 = np.where(im_new>np.median(im_new), 1,0)
_,drift = reg(im1, im2, params)
roi.append(drift)
except StopIteration:
pass
finally:
del im
return np.array(roi)
def interpol_drift(drift, params):
"""Returns linearly interpolated ROI.
This uses drift calculation where drift comes from adjacent reference frame."""
x = np.cumsum(drift[:,1])
y = np.cumsum(drift[:,0])
r = np.zeros((params['nof'],2))
dr = params['nof']%params['chunk']
for cnt in xrange(1,len(r)-dr):
index = float(cnt)/(params["chunk"])
i = int(index)+1
vy, vx = y[i]-y[i-1], x[i]-x[i-1]
#r[cnt] = (index%1*vy)+y[i-1],(index%1*vx)+x[i-1]
r[cnt] = y[i-1]+(index%1)*vy,x[i-1]+(index%1)*vx
if cnt == len(r)-dr-1:
#this deals with leftover interval if images%chunk!=0
for rest in xrange(1,dr+1):
index = float(rest)/dr+1
vy, vx = y[i]-y[i-1], x[i]-x[i-1]
r[cnt+rest] = y[i-1]+(index%1*vy),x[i-1]+(index%1*vx)
return r
##===================================================#
# feature detection for neuron physiological imaging
## ==================================================#
def fluorescence(params, roi):
"""finds a neuron from images using thresholding
in a region of interest."""
images = read_sequentially(params)
values,locations = [],[]
try:
cnt = 0
imgs = ndimage.shift(images.next(), roi[cnt], mode="wrap")
cms_old = [params['y0'],params['x0']]
#print "cms_old is ",cms_old
val_old = []
y0, x0 = cms_old
height, width = imgs.shape
cnt += 1
while True:
y1, x1, fluor, bg = similarity3(imgs, cms_old,[y0,x0], params)
#implement a short memory of neuron position
val_old.append([y1, x1])
y0 = np.average([v[0] for v in val_old[-10:]])
x0 = np.average([v[1] for v in val_old[-10:]])
values.append([fluor, bg])
locations.append([y1-roi[cnt-1][0],x1+params["cropx"][0]-roi[cnt-1][1]])
imgs = ndimage.shift(images.next(), roi[cnt], mode="wrap")
cnt += 1
except StopIteration:
pass
finally:
del images
return np.array(values), np.array(locations)
def similarity3(im1, cms,old_coor, params):
"""Calculates fluorescence of neuron by thresholding."""
bgsize = params["bgsize"]
part1 = im1[max(0,cms[0]-bgsize):cms[0]+bgsize, max(0,cms[1]-bgsize):cms[1]+bgsize]
offsety, offsetx = max(0,cms[0]-bgsize), max(0,cms[1]-bgsize)
height, width = part1.shape
y0,x0 = old_coor #previous coords
thresh = np.sort(part1, axis=None)[-int(params["thresh_pump"]*height*width)]
#print "threshold is", thresh
mask = np.where(part1 > thresh, 1, 0)
mask = ndimage.binary_opening(mask,structure = np.ones((2,2)))
mask = ndimage.binary_closing(mask)
label_im, nb_labels = ndimage.label(mask)
centroids = ndimage.measurements.center_of_mass(part1, label_im, xrange(1,nb_labels+1))
dist = []
for index, coord in enumerate(centroids):
y,x= coord
dist.append((y-y0+offsety)**2 + (x-x0+offsetx)**2)
if min(dist)>2*params["max_movement"]**2:
print dist, y0,x0, offsety, offsetx,
y,x = y0-offsety,x0-offsetx
radius = params["roisize"]
neuron = part1[max(0,y-radius):y+radius,max(0,x-radius):x+radius,]
value = np.ma.average(np.sort(neuron, axis=None)[-20:])
else:
loc = np.argmin(dist)
y,x = centroids[loc]
remove_pixel = np.where(label_im ==loc+1,0,1)
neuron = np.ma.masked_array(part1, remove_pixel)
value = np.ma.average(neuron)
try:
radius = params["roisize"]
mask1 = np.zeros(part1.shape, dtype=bool)
mask1[max(0,y-radius):y+radius,max(0,x-radius):x+radius,] = True
bg_mask = np.ma.mask_or(mask,mask1)
bg = np.ma.masked_array(part1, bg_mask)
bg_level = np.ma.average(bg)
except IndexError:
y,x=y0,x0
value=0
bg_level=0
return y+offsety, x+offsetx, value, bg_level
##===================================================#
# Main
## ==================================================#
def warp_detector(params):
##===================================================#
# Translation correction
## ==================================================#
drift = find_roi(params)
drift = interpol_drift(drift, params)
print "done with drift"
sys.stdout.flush()
##===================================================#
# detect pumping
## ==================================================#
coords = fluorescence(params, drift)
time = np.arange(params["start"], params["start"]+len(coords),1)
out_data = zip(time,coords[:,0], coords[:,1],coords[:,2])
print "Analysis of: ",params["start"], params["end"]
##===================================================#
# write results and movie
## ==================================================#
if len(coords) > 0:
outputstring = "%s_%i_%i"%(params["basename"],params["start"],params["end"]-1)
pump_writer.write_data(params["outdir"], outputstring+"_kymo", out_data, 4)
images = read_sequentially(params)
fig = plt.figure(params["start"]+1) #make unique figures needed for parallelization
ax1 = fig.add_subplot(211)
pump_writer.make_kymograph(images, params, diff=False, roi=drift)
ax1.plot(cms[:,0]+drift[:,0])
ax1.set_xlim([0,len(cms)])
ax2 = fig.add_subplot(212)
ax2.plot(coords[:,0])
ax2.plot(coords[:,1])
ax2.set_xlim([0,len(coords)])
ax2.set_ylim(ax2.get_ylim()[::-1])
fig.savefig(params["outdir"]+"/"+outputstring+"_kym.png")