-
Notifications
You must be signed in to change notification settings - Fork 89
/
normcorre_batch_even.m
330 lines (307 loc) · 12.4 KB
/
normcorre_batch_even.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
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
function [M_final,shifts_g,template,options,col_shift] = normcorre_batch_even(Y,options,template)
% online motion correction through DFT subpixel registration
% Based on the dftregistration.m function from Manuel Guizar and Jim Fienup
% INPUTS
% Y: Input data, can be already loaded in memory as a 3D
% tensor, a memory mapped file, or a pointer to a tiff stack
% options: options structure for motion correction (optional, rigid registration is performed if not provided)
% template: provide template (optional)
% OUTPUTS
% M_final: motion corrected data
% shifts_up: upsampled shifts
% shifts: originally calculated shifts
% template: calculated template
%% first determine filetype
if isa(Y,'char')
[~,~,ext] = fileparts(Y);
ext = ext(2:end);
if strcmpi(ext,'tif') || strcmpi(ext,'tiff')
tiffInfo = imfinfo(Y);
filetype = 'tif';
T = length(tiffInfo);
sizY = [tiffInfo(1).Height,tiffInfo(1).Width,T];
elseif strcmpi(ext,'mat')
filetype = 'mem';
Y = matfile(Y,'Writable',true);
details = whos(Y);
var_sizes = [details.bytes];
[~,var_ind] = max(var_sizes);
var_name = details(var_ind).name;
sizY = size(Y,var_name);
T = sizY(end);
elseif strcmpi(ext,'hdf5') || strcmpi(ext,'h5')
filetype = 'hdf5';
fileinfo = hdf5info(Y);
data_name = fileinfo.GroupHierarchy.Datasets.Name;
sizY = fileinfo.GroupHierarchy.Datasets.Dims;
T = sizY(end);
elseif strcmpi(ext,'raw')
filetype = 'raw';
fid = fopen(Y);
FOV = [options.d1,options.d2];
bitsize = options.bitsize;
imsize = FOV(1)*FOV(2)*bitsize; % Bit size of single frame
current_seek = ftell(fid);
fseek(fid, 0, 1);
file_length = ftell(fid);
fseek(fid, current_seek, -1);
T = file_length/imsize;
sizY = [FOV,T];
fclose(fid);
elseif strcmpi(ext,'avi')
filetype = 'avi';
sizY = size(read_file(Y));
FOV = sizY(1:2);
T = sizY(end);
end
elseif isobject(Y)
filetype = 'mem';
var_name = 'Y';
sizY = size(Y,var_name);
T = sizY(end);
else % array loaded in memory
filetype = 'mat';
sizY = size(Y);
T = sizY(end);
end
nd = length(sizY)-1; % determine whether imaging is 2d or 3d
sizY = sizY(1:nd);
otherdims = repmat({':'},1,nd);
%% set default parameters if not present
if ~exist('options','var') || isempty(options)
options = NoRMCorreSetParms('d1',sizY(1),'d2',sizY(2));
if nd > 2; options.d3 = sizY(3); end
end
memmap = options.memmap;
grid_size = options.grid_size;
mot_uf = options.mot_uf;
min_patch_size = options.min_patch_size;
min_diff = options.min_diff;
overlap_pre = options.overlap_pre;
overlap_post = options.overlap_post;
upd_template = options.upd_template;
bin_width = options.bin_width;
buffer_width = options.buffer_width;
max_dev_g = options.max_dev;
init_batch = options.init_batch;
us_fac = options.us_fac;
method = options.method;
filename = options.mem_filename;
iter = options.iter;
add_value = options.add_value;
max_shift = options.max_shift;
if strcmpi(options.boundary,'nan')
fill_value = NaN;
else
fill_value = add_value;
end
print_msg = options.print_msg;
while mod(T,bin_width) == 1
if T == 1
error('Movie appears to have only one frame. Use the function normcorre instead')
end
bin_width = bin_width + 1;
end
%% first check for offset due to bi-directional scanning
if options.correct_bidir && isempty(options.col_shift)
col_shift = correct_bidirectional_offset(Y,options.nFrames,options.bidir_us);
elseif ~isempty(options.col_shift)
col_shift = options.col_shift;
else
col_shift = 0;
end
options.col_shift = col_shift;
if col_shift
if print_msg; fprintf('Offset %1.1d pixels due to bidirectional scanning detected. \n',col_shift); end
if strcmpi(options.shifts_method,'fft')
options.shifts_method = 'cubic';
if print_msg; fprintf('Cubic shifts will be applied. \n'); end
end
end
%% read initial batch and compute template
init_batch = min(T,init_batch);
interval = ceil(T/2-init_batch/2+1):floor(T/2+init_batch/2);
switch filetype
case 'tif'
Y_temp = read_file(Y,interval(1),init_batch,[],tiffInfo);
case 'hdf5'
Y_temp = read_file(Y,interval(1),init_batch);
case 'avi'
Y_temp = read_file(Y,interval(1),init_batch);
case 'mem'
Y_temp = Y.(var_name)(otherdims{:},interval);
case 'mat'
Y_temp = Y(otherdims{:},interval);
case 'raw'
Y_temp = read_raw_file(Y,interval(1),init_batch,FOV,bitsize);
end
data_type = class(Y_temp);
Y_temp = single(Y_temp);
use_proj = true;
if nargin < 3 || isempty(template)
if print_msg; fprintf('Registering the first %i frames just to obtain a good template....',init_batch); end
template_in = median(Y_temp,nd+1)+add_value;
fftTemp = fftn(template_in);
for t = 1:size(Y_temp,nd+1)
if nd == 2
[~,Greg] = dftregistration_min_max(fftTemp,fftn(Y_temp(:,:,t)),us_fac,-max_shift,max_shift,options.phase_flag);
end
if nd == 3
[~,Greg] = dftregistration_min_max_3d(fftTemp,fftn(Y_temp(:,:,:,t)),us_fac,-max_shift,max_shift,options.phase_flag);
end
M_temp = real(ifftn(Greg));
template_in = template_in*(t-1)/t + M_temp/t;
end
template_in = template_in + add_value;
if print_msg; fprintf('..done. \n'); end
else
template_in = single(template + add_value);
end
[d1,d2,d3,~] = size(Y_temp);
if nd == 2; d3 = 1; end
%% setup grids for patches
dim = [d1,d2,d3];
patches = construct_grid_even(grid_size,overlap_pre,dim,min_diff);
shifts_g = struct('shifts',cell(T,1),'shifts_up',cell(T,1),'diff',cell(T,1));
template_patches = split_frame(template_in,patches);
patch_size = size(template_patches);
n_patches = [length(unique(patches(:,1))),length(unique(patches(:,3))),length(unique(patches(:,5)))];
%%
%maxNumCompThreads(1);
temp_mat = template_in;
fftTemp =fft(fft(fft(template_patches,[],1),[],2),[],3);
fftTempMat = fftn(temp_mat);
if ~strcmpi(options.output_type,'mat')
options.mem_batch_size = max(min(round(options.mem_batch_size/bin_width)*bin_width,T),1);
mem_buffer = squeeze(zeros([dim,options.mem_batch_size],'single'));
end
switch lower(options.output_type)
case 'mat'
M_final = zeros([sizY,T],data_type);
case 'memmap'
M_final = matfile(filename,'Writable',true);
if nd == 2; M_final.Y(d1,d2,T) = zeros(1,data_type); end
if nd == 3; M_final.Y(d1,d2,d3,T) = zeros(1,data_type); end
M_final.Yr(d1*d2*d3,T) = zeros(1,data_type);
case {'hdf5','h5'}
if exist(options.h5_filename,'file')
[pathstr,fname,ext] = fileparts(options.h5_filename);
new_filename = fullfile(pathstr,[fname,'_',datestr(now,30),ext]);
warning_msg = ['File ',options.h5_filename,'already exists. Saving motion corrected file as',new_filename];
warning('%s',warning_msg);
options.h5_filename = new_filename;
end
M_final = options.h5_filename;
if nd == 2
h5create(options.h5_filename,['/',options.h5_groupname],[d1,d2,Inf],'Chunksize',[d1,d2,options.mem_batch_size],'Datatype',data_type);
elseif nd == 3
h5create(options.h5_filename,['/',options.h5_groupname],[d1,d2,d3,Inf],'Chunksize',[d1,d2,d3,options.mem_batch_size],'Datatype',data_type);
end
case {'tif','tiff'}
M_final = ['motion corrected file has been saved as ', options.tiff_filename];
opts_tiff.append = true;
opts_tiff.big = true;
if nd == 3
error('Saving volumetric tiff stacks is currently not supported. Use a different filetype');
end
otherwise
error('This filetype is currently not supported')
end
cnt_buf = 0;
if print_msg; fprintf('Template initialization complete. Now registering all the frames with new template. \n'); end
%%
prevstr = [];
for it = 1:iter
for t = 1:bin_width:T
switch filetype
case 'tif'
Ytm = single(read_file(Y, t, min(t+bin_width-1,T)-t+1, [], tiffInfo));
case 'avi'
Ytm = single(read_file(Y, t, min(t+bin_width-1,T)-t+1));
case 'hdf5'
Ytm = single(h5read(Y,data_name,[ones(1,nd),t],[sizY(1:nd),min(t+bin_width-1,T)-t+1]));
case 'mem'
Ytm = single(Y.(var_name)(otherdims{:},t:min(t+bin_width-1,T)));
case 'mat'
Ytm = single(Y(otherdims{:},t:min(t+bin_width-1,T)));
case 'raw'
Ytm = single(read_raw_file(Y,t,min(t+bin_width-1,T)-t+1,FOV,bitsize));
end
Mf = zeros(size(Ytm),data_type);
lY = size(Ytm,nd+1);
shifts = struct('shifts',cell(lY,1),'shifts_up',cell(lY,1),'diff',cell(lY,1));
for i = lY:-1:1
Yt = Ytm(otherdims{:},i);
future_results(i) = parfeval(@register_frame, 2, Yt,fftTempMat,fftTemp,patches,options);
end
for i = 1:lY
[idx, shifts_temp, Mf_temp] = fetchNext(future_results);
shifts(idx).shifts = shifts_temp;
shifts(idx).shifts_up = shifts_temp;
Mf(otherdims{:},idx) = Mf_temp;
end
shifts_g(t:min(t+bin_width-1,T)) = shifts;
if ~strcmpi(options.output_type,'mat')
rem_mem = rem(t+lY-1,options.mem_batch_size);
if rem_mem == 0; rem_mem = options.mem_batch_size; end
mem_buffer(otherdims{:},rem_mem-lY+1:rem_mem) = cast(Mf,data_type);
end
if it == iter
switch lower(options.output_type)
case 'mat'
M_final(otherdims{:},t:min(t+bin_width-1,T)) = cast(Mf,data_type);
case 'memmap'
if rem_mem == options.mem_batch_size || t+lY-1 == T
M_final.Y(otherdims{:},t+lY-rem_mem:t+lY-1) = mem_buffer(:,:,1:rem_mem);
M_final.Yr(:,t+lY-rem_mem:t+lY-1) = reshape(mem_buffer(1:d1*d2*d3*rem_mem),d1*d2*d3,rem_mem);
end
case {'hdf5','h5'}
if rem_mem == options.mem_batch_size || t+lY-1 == T
if nd == 2; h5write(options.h5_filename,['/',options.h5_groupname],mem_buffer(:,:,1:rem_mem),[ones(1,nd),t+lY-rem_mem],[sizY(1:nd),rem_mem]); end
if nd == 3; h5write(options.h5_filename,['/',options.h5_groupname],mem_buffer(:,:,:,1:rem_mem),[ones(1,nd),t+lY-rem_mem],[sizY(1:nd),rem_mem]); end
end
case {'tif','tiff'}
if rem_mem == options.mem_batch_size || t+lY-1 == T
saveastiff(cast(mem_buffer(:,:,1:rem_mem),data_type),options.tiff_filename,opts_tiff);
end
end
end
if print_msg
str = sprintf('%d out of %d frames registered, iteration %d out of %d..', t+lY-1, T, it, iter);
refreshdisp(str, prevstr, t);
prevstr=str;
end
% update template
if upd_template
cnt_buf = cnt_buf + 1;
if strcmpi(method{2},'mean')
new_temp = nanmean(Mf,nd+1);
elseif strcmpi(method{2},'median')
new_temp = nanmedian(x,nd+1);
end
if any(isnan(new_temp(:)))
new_temp(isnan(new_temp)) = template(isnan(new_temp));
end
if strcmpi(method{1},'mean')
cnt = t/bin_width + 1;
temp_mat = temp_mat*(cnt-1)/cnt + new_temp/cnt;
elseif strcmpi(method{1},'median')
ind_buffer = mod(cnt_buf,buffer_width) + buffer_width*(~mod(cnt_buf,buffer_width));
buffer_med(otherdims{:},ind_buffer) = new_temp;
temp_mat = nanmedian(buffer_med,nd+1);
end
template_patches = split_frame(temp_mat,patches);
fftTemp = fft(fft(fft(template_patches,[],1),[],2),[],3);
fftTempMat = fftn(temp_mat);
end
end
if it == iter
template = temp_mat - add_value;
end
if memmap
M_final.shifts = shifts_g;
M_final.template = template;
end
end
if print_msg; fprintf('\n'); end
maxNumCompThreads('automatic');