-
Notifications
You must be signed in to change notification settings - Fork 2
/
cell_coloc_3D.m
310 lines (245 loc) · 8.77 KB
/
cell_coloc_3D.m
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
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
function cell_coloc_3D
%% Adam Tyson | 2018-03-26 | [email protected]
% loads C0 image (e.g. DAPI), displays, and allows manual seg of each object
% each cell is then segmented, and intensity of a secondary marker C2
% (within C0) is measured.
%% TO DO
% add option to only analyse certain images
% removal of any big (double) cells
% update README for new outputs
% remove multiple (slow) calls to bwconvhull
% only load subsampled data, rather than loading all first, to not use
%% IMPROVE SEGMENTATION
vars=getVars;
tic
cd(vars.directory)
files=dir('*C0.tif'); % all tif's in this folder
numImages=length(files);
imCount=0;
% manually segment objects (e.g spheroids)
for file=files' % go through all images
imCount=imCount+1;
C0file{imCount}=file.name;
% Load images and separate objects
tmpIm=loadFile(C0file{imCount});
rawC0{imCount}=imresize(tmpIm, vars.zScale);
[bin_C0{imCount}, objNum{imCount}] = manSeg(rawC0{imCount});
end
objInf=cell(2, imCount);
objInf{2,1}= "Number of cells per object";
objInf{3,1}= "Total volume of object (nuclei only)";
objInf{4,1}= "Total volume of object (convex bounding)";
objInf{5,1}= "Object density";
objInf{6,1}= "Mean marker intensity per object";
objInf{7,1}= "Mean cell size per object";
% Load C2 and analyse each object
f = waitbar(0,'1','Name','Analysing images...');
count=0;
for im=1:imCount
count=count+1;
waitbar((count-1)/numImages,f,strcat("Analysing Image: ", num2str(count)))
C2file{im} = replace(C0file{im},'C0','C2');
tmpIm=loadFile(C2file{im});
rawC2=imresize(tmpIm, vars.zScale);
clear tmpIm
[rawC0_ind, rawC2_ind] = maskObj(rawC0{im}, rawC2,...
bin_C0{im}, objNum{im}); % mask images
segC0=segment3D(rawC0_ind, vars); % segment
% mean C2 fluro per cell, per object
[C0sizes, C2means, objBoundVol]=indv_cell_coloc(segC0, rawC2_ind);
%% summary results
[~, nametmp,~] = fileparts(C0file{im});
objInf{1, im+1}= strcat("Image_", nametmp);
objInf{2, im+1} = cellfun(@(x) max(x(:)), segC0); % no cells per obj
objInf{3, im+1} = cellfun(@(x) nnz(x>0), segC0); % vol obj (cells)
objInf{4, im+1} = objBoundVol; % vol obj (convex)
objInf{5, im+1} = objInf{3, im+1}./objInf{4, im+1}; % density
% get mean vals
objC2Means=[];
objSizeMeans=[];
for obj=1:objNum{im}
objC2Means=[objC2Means mean(cell2mat(C0sizes(obj,:)))];
objSizeMeans=[objSizeMeans mean(cell2mat(C2means(obj,:)))];
end
objInf{6, im+1} = objC2Means;
objInf{7, im+1} = objSizeMeans;
if strcmp(vars.plot, 'Yes')
res_vis(C2means, vars, C0file{im});
end
if strcmp(vars.saveSegmentation, 'Yes')
saveSegmentation(objNum, rawC0_ind, rawC2_ind, segC0,...
im, C0file, C2file)
end
if strcmp(vars.savecsv, 'Yes')
save_raw_res(C0file, C0sizes, C2means, im)
end
end
if strcmp(vars.savecsv, 'Yes')
save_summary_res(objInf)
end
delete(f)
toc
end
%% Internal functions
function save_summary_res(objectInfo)
csvname="summary_results.csv";
results_Table=cell2table(objectInfo);
writetable(results_Table, csvname, 'WriteVariableNames', 0)
end
function save_raw_res(C0file, C0sizes, C2means, im)
% tidy up
%% mean marker intensities
[~, nametmp,~] = fileparts(C0file{im});
csvname = ['marker_mean_intensity_' nametmp '.csv'];
% add labels
sze=size(C2means);
blankY=cell(sze(1),1);
blankX=cell(1, sze(2)+1);
C2means=[blankY C2means];
C2means=[blankX; C2means];
for cellnum=1:sze(2)
C2means{1, cellnum+1}=strcat("Cell_", num2str(cellnum));
end
for obj=1:sze(1)
C2means{obj+1,1}=strcat("Object_", num2str(obj));
end
results_Table=cell2table(C2means);
writetable(results_Table, csvname, 'WriteVariableNames', 0)
%% mean cell sizes
csvname2 = ['cell_sizes_' nametmp '.csv'];
% add labels
sze=size(C0sizes);
blankY=cell(sze(1),1);
blankX=cell(1, sze(2)+1);
C0sizes=[blankY C0sizes];
C0sizes=[blankX; C0sizes];
for cellnum=1:sze(2)
C0sizes{1, cellnum+1}=strcat("Cell_", num2str(cellnum));
end
for obj=1:sze(1)
C0sizes{obj+1,1}=strcat("Object_", num2str(obj));
end
results_Table2=cell2table(C0sizes);
writetable(results_Table2, csvname2, 'WriteVariableNames', 0)
end
function saveSegmentation(objNum, rawC0_ind, rawC2_ind, segC0,...
im, C0file, C2file)
for obj=1:objNum{im}
C0raw_tmp=rawC0_ind{obj};
C2raw_tmp=rawC2_ind{obj};
C0seg_tmp=segC0{obj};
outC0_raw=['raw_obj_' num2str(obj) '_' C0file{im}];
outC2_raw=['raw_obj_' num2str(obj) '_' C2file{im}];
outC0_seg=['seg_obj_' num2str(obj) '_' C0file{im}];
for frame=1:size(C0raw_tmp,3)
imwrite(C0raw_tmp(:,:,frame),outC0_raw,...
'tif','WriteMode', 'append', 'compression', 'none');
imwrite(C2raw_tmp(:,:,frame),outC2_raw,...
'tif', 'WriteMode', 'append', 'compression', 'none');
imwrite(C0seg_tmp(:,:,frame),outC0_seg,...
'tif', 'WriteMode', 'append', 'compression', 'none');
end
end
end
function image=loadFile(file)
disp(['Loading: ' file])
info = imfinfo(file);
numZ = numel(info);
image=uint16(zeros(info(1).Height, info(1).Width, numZ)); %initalise
for k = 1:numZ
image(:,:,k) = imread(file, k, 'Info', info); % load frame by frame
end
end
function [binaryImages, objNum]=manSeg(image)
scrsz = get(0,'ScreenSize');
imSize=size(image);
dispScale=(scrsz(4)/imSize(1))*0.8;
screenSize=[10 10 dispScale*imSize(2) dispScale*imSize(1)];
% Plot intensity projection.
image_max = max(image, [], 3);
figure('position', screenSize,'Name','Manually segment objects')
imagesc(image_max)
colormap gray
continueSeg=1;
objNum=1;
while continueSeg==1
hFH = imfreehand(); % manually segment
tmpBin = hFH.createMask(); % make binary image
repSeg = questdlg('Redo last segmentation?',...
'Error catch','Yes','No','No'); % yes, no and default
if strcmp(repSeg, 'No')
binaryImages(:,:,objNum) = tmpBin;
finSeg = questdlg('All objects segmented?',...
'Error catch','Yes','No','No');
if strcmp(finSeg, 'No')
objNum = objNum+1;
elseif strcmp(finSeg, 'Yes')
continueSeg = 0;
close all
end
end
end
end
function imageCrop=deleteZeros(image)
image_max = max(image, [], 3);
for z=1:size(image,3)
imagetmp=image(:,:,z);
imagetmp( all(~image_max,2), :) = []; % remove zero rows
imagetmp( :, all(~image_max,1)) = []; % remove zero columns
imageCrop(:,:,z)=imagetmp;
end
end
function [C0_indiv, C2_indiv] = maskObj(C0_image,...
C2_image, binaryImages, objNum)
C0_indiv = cell(objNum, 1) ;
C2_indiv = cell(objNum, 1) ;
for object=1:objNum
for z=1:size(C0_image,3)
C0_indv_tmp(:,:,z)=C0_image(:,:,z).*...
uint16(binaryImages(:,:,object));
C2_indv_tmp(:,:,z)=C2_image(:,:,z).*...
uint16(binaryImages(:,:,object));
end
imageCrop_C0=deleteZeros(C0_indv_tmp);
C0_indiv{object}=imageCrop_C0;
imageCrop_C2=deleteZeros(C2_indv_tmp);
C2_indiv{object}=imageCrop_C2;
end
end
function vars=getVars
vars.directory = uigetdir('', 'Choose directory containing images');
vars.savecsv = questdlg('Save results as .csv?', ...
'Exporting', ...
'Yes', 'No', 'Yes');
vars.plot = questdlg('Plot individual heat maps? ', ...
'Plotting', ...
'Yes', 'No', 'No');
vars.saveSegmentation= questdlg('Save segmentation as.tif?', ...
'Saving segmentation', ...
'Yes', 'No', 'No');
vars.edgeRem= questdlg('Remove edge objects?', ...
'Clear edges', ...
'Yes', 'No', 'Yes');
prompt = {'Segmentation threshold (a.u.):',...
'Smoothing width (pixels):',...
'Maximum hole size to fill (pixels):',...
'Largest false cell to remove (pixels):',...
'Watershed threshold (a.u.):',...
'Voxel size - XY (um):',...
'Voxel size - Z (um):'};
dlg_title = 'Analysis variables';
num_lines = 1;
defaultans = {'1.4', '2', '50', '30', '3.5', '0.065', '0.34'};
answer = inputdlg(prompt,dlg_title,num_lines,defaultans);
vars.threshScale=str2double(answer{1});%change sensitivity of threshold
vars.smoothSigma=str2double(answer{2});% smoothing kernel
vars.holeSize=str2double(answer{3});% largest hole to fill
vars.noiseRem=str2double(answer{4}); % smallest obj to remove
vars.localMaxThresh=str2double(answer{5});% ws int marker threshold
vars.xySamp=str2double(answer{6});% vox size
vars.zSamp=str2double(answer{7});% vox size
vars.zScale=vars.xySamp/vars.zSamp;
vars.stamp=num2str(fix(clock)); % date and time
vars.stamp(vars.stamp==' ') = '';%remove spaces
vars.fontSize=14;
end