forked from stabinska/IDEAL
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathIDEALfitIVIM.m
275 lines (232 loc) · 9.58 KB
/
IDEALfitIVIM.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
function [FitResults,FitQuality,P,ROIstat] = IDEALfitIVIM(DataNii,P,MaskNii,ROINii)
%% Function IDEALfit
%
% Fit a bi-exponential or tri-exponential model to DWI data using the
% Iterative Downsampling Adaptive Least-squares (IDEAL) approach
%
%
% Input: Data(x_pos, y_pos, z_pos, b-value) - 4D-DWI data (*.nii.gz)
% P - structure with experiment and fit parameters
% MaskNii - mask (e.g. both kidneys) (*.nii.gz)
% ROINii - regions of interest (e.g. cortex, medulla, pelvis) (*.nii.gz) as cell
%
%
% Output: FitResults - structure with fit results
% FitQuality - structure with goodnes-of-fit measures (SSE, R2, adjR2, stdError, CVs)
% P - updated structure with experiment and fit parameters
%
%
% Authors: Julia Stabinska (jstabin3@jhmi.edu)
% Helge Jörn Zöllner (hzoelln2@jhmi.edu)
% Thomas Andreas Thiel (thomas.thiel@hhu.de)
%
%% Parse input arguments
if nargin < 3
error('You need to supply at least a 3D-DWI data matrix as *.nii.gz, a struct with settings (see IDEAL script), and a mask volume of the volume of interest as *.nii.gz');
end
%% Load data, mask ,and ROIs
data_nifti = double(rot90(niftiread(DataNii)));
Data = squeeze(data_nifti(:,:,P.slice,:));
Mask_nifti = double(rot90(niftiread(MaskNii)));
Mask = Mask_nifti(:,:,P.slice);
Mask_nan = Mask;
Mask_nan(Mask == 0) = NaN;
if nargin > 3
ROIs{1} = double(rot90(niftiread(MaskNii)));
ROIs{1} = ROIs{1}(:,:,P.slice);
ROIs{1}(ROIs{1}==0) =NaN;
for rois = 2 : length(ROINii)+1
ROIs{rois} = double(rot90(niftiread(ROINii{rois-1})));
ROIs{rois}(ROIs{rois}==0) =NaN;
end
else
ROIs{1} = double(rot90(niftiread(MaskNii)));
ROIs{1} = ROIs{1}(:,:,P.slice);
ROIs{1}(ROIs{1}==0) =NaN;
sprintf('No ROI supplied. We will do statistics on the VOI mask.')
end
%% Perform IDEAL fitting
clearvars a b c d e f fitresults gof output
Data_masked = Data.*Mask_nan;
tStart = tic;
for res = 1 : size(P.Dims_steps, 1)
fprintf('Downsampling step no: %s\n', num2str(res));
a = zeros(P.Dims_steps(res,1), P.Dims_steps(res,2));
b = zeros(P.Dims_steps(res,1), P.Dims_steps(res,2));
c = zeros(P.Dims_steps(res,1), P.Dims_steps(res,2));
d = zeros(P.Dims_steps(res,1), P.Dims_steps(res,2));
e = zeros(P.Dims_steps(res,1), P.Dims_steps(res,2));
f = zeros(P.Dims_steps(res,1), P.Dims_steps(res,2));
if res < size(P.Dims_steps, 1)
Mask_res = imresize(Mask, [P.Dims_steps(res,2) P.Dims_steps(res,1)],'bilinear');
Data_res = zeros(P.Dims_steps(res,1), P.Dims_steps(res,2),16);
for i = 1:numel(P.b_values)
Data_res(:,:,i) = imresize(squeeze(Data(:, :, i)),[P.Dims_steps(res,2) P.Dims_steps(res,1)],'bilinear');
end
Mask_res = abs(Mask_res);
Mask_res(Mask_res < 0.025) = 0;
else
Data_res = Data_masked;
Mask_res = Mask;
end
FitResults = cell(size(Mask_res, 1), size(Mask_res, 2));
gof = FitResults;
output = FitResults;
for x = 1 : size(Mask_res, 1)
for y = 1: size(Mask_res, 2)
op.Display = 'Off';
op.Algorithm = 'Trust-Region';
op.MaxIter = 600;
if res == 1
op.Lower = P.op.Lower;
op.StartPoint = P.op.StartPoint;
op.Upper = P.op.Upper;
else
% f_inter f_fast D_slow D_inter D_fast S0
op.Lower = [a_res(x,y)*(1-P.Tol(1)) b_res(x,y)*(1-P.Tol(1)) c_res(x,y)*(1-P.Tol(2)) d_res(x,y)*(1-P.Tol(2)) e_res(x,y)*(1-P.Tol(2)) f_res(x,y)*(1-P.Tol(3))];
op.StartPoint = [a_res(x,y) b_res(x,y) c_res(x,y) d_res(x,y) e_res(x,y) f_res(x,y)];
op.Upper = [a_res(x,y)*(1+P.Tol(1)) b_res(x,y)*(1+P.Tol(1)) c_res(x,y)*(1+P.Tol(2)) d_res(x,y)*(1+P.Tol(2)) e_res(x,y)*(1+P.Tol(2)) f_res(x,y)*(1+P.Tol(3))];
end
if Mask_res(x,y)
if ~isnan(Data_res(x,y,1))
[FitResults{x,y}, gof{x,y}, output{x,y}] = TriexpFit(P.b_values, squeeze(Data_res(x,y,:))', op);
a(x,y) = FitResults{x,y}.a;
b(x,y) = FitResults{x,y}.b;
c(x,y) = FitResults{x,y}.c;
d(x,y) = FitResults{x,y}.d;
e(x,y) = FitResults{x,y}.e;
f(x,y) = FitResults{x,y}.f;
end
end
end
end
% Interpolate parameters
if res < size(P.Dims_steps, 1)
a_res = imresize(a,[P.Dims_steps(res+1,2) P.Dims_steps(res+1,1)], 'bilinear');
b_res = imresize(b,[P.Dims_steps(res+1,2) P.Dims_steps(res+1,1)], 'bilinear');
c_res = imresize(c,[P.Dims_steps(res+1,2) P.Dims_steps(res+1,1)], 'bilinear');
d_res = imresize(d,[P.Dims_steps(res+1,2) P.Dims_steps(res+1,1)], 'bilinear');
e_res = imresize(e,[P.Dims_steps(res+1,2) P.Dims_steps(res+1,1)], 'bilinear');
f_res = imresize(f,[P.Dims_steps(res+1,2) P.Dims_steps(res+1,1)], 'bilinear');
end
end
P.time = toc(tStart);
%% Extract the parameters maps
f_slow = nan(size(Mask));
f_interm = nan(size(Mask));
f_fast = nan(size(Mask));
D_slow = nan(size(Mask));
D_interm = nan(size(Mask));
D_fast = nan(size(Mask));
S_0 = nan(size(Mask));
SSE = nan(size(Mask));
Rsq = nan(size(Mask));
Dfe = nan(size(Mask));
AdjRsq = nan(size(Mask));
RMSE = nan(size(Mask));
Residuals = nan(size(Mask, 1), size(Mask, 2), 16);
for kx = 1:size(Mask, 1)
for ky = 1:size(Mask, 2)
if ~isempty(FitResults{kx,ky})
f_slow(kx, ky) = 1 - FitResults{kx,ky}.a - FitResults{kx,ky}.b;
f_interm(kx, ky) = FitResults{kx,ky}.a;
f_fast(kx, ky) = FitResults{kx,ky}.b;
D_slow(kx, ky) = FitResults{kx,ky}.c;
D_interm(kx, ky) = FitResults{kx,ky}.d;
D_fast(kx, ky) = FitResults{kx,ky}.e;
S_0(kx, ky) = FitResults{kx,ky}.f;
FitQuality.SSE(kx, ky) = gof{kx,ky}.sse;
FitQuality.Rsq(kx, ky) = gof{kx,ky}.rsquare;
FitQuality.Dfe(kx, ky) = gof{kx,ky}.dfe;
FitQuality.AdjRsq(kx, ky) = gof{kx,ky}.adjrsquare;
FitQuality.RMSE(kx, ky) = gof{kx,ky}.rmse;
FitQuality.Residuals(kx, ky, :) = output{kx,ky}.residuals;
end
end
end
%% Plot the parameter maps
[~,file_name,~] = fileparts(DataNii);
if P.plot
figure('Visible','on')
subplot(3,3,1)
imagesc(f_slow);
caxis(gca,[0 1])
title('f_{slow}')
colormap gray;
axis off;
subplot(3,3,2)
imagesc(f_interm);
caxis(gca,[0 1])
title('f_{interm}')
colormap gray;
axis off;
subplot(3,3,3)
imagesc(f_fast);
caxis(gca,[0 1])
title('f_{fast}')
colormap gray;
axis off;
subplot(3,3,4)
imagesc(D_slow);
title('D_{slow}')
colormap gray;
axis off;
subplot(3,3,5)
imagesc(D_interm);
title('D_{interm}')
colormap gray;
axis off;
subplot(3,3,6)
imagesc(D_fast);
title('D_{fast}')
colormap gray;
axis off;
subplot(3,3,7)
imagesc(S_0);
title('S_{0,fit}');
colormap gray;
axis off;
subplot(3,3,8)
imagesc(squeeze(Data(:, :, 1)));
title('S_{0}');
colormap gray;
axis off;
if ~exist(P.outputFolder)
mkdir(P.outputFolder);
end
fignm_param = sprintf([P.outputFolder,'%sIDEALfit_%s_steps_%s_param.fig'], filesep, file_name, num2str(size(P.Dims_steps, 1)));
savefig(gcf, fignm_param);
close(gcf);
end
%% Perform ROI-based analysis
ROIstat.ROIname = cell(1,length(ROIs));
IVIMPars = {'f_slow','f_interm','f_fast','D_slow','D_interm','D_fast','S_0'};
for par = 1 : length(IVIMPars)
ROIstat.(IVIMPars{par}).mean = zeros(1,length(ROIs));
ROIstat.(IVIMPars{par}).median = zeros(1,length(ROIs));
ROIstat.(IVIMPars{par}).std = zeros(1,length(ROIs));
ROIstat.(IVIMPars{par}).CV = zeros(1,length(ROIs));
ROIstat.(IVIMPars{par}).iqr = zeros(1,length(ROIs));
ROIstat.(IVIMPars{par}).q1 = zeros(1,length(ROIs));
ROIstat.(IVIMPars{par}).q3 = zeros(1,length(ROIs));
end
for rois = 1 : length(ROIs)
if rois == 1
[~,name,~] = fileparts(MaskNii);
else
[~,name,~] = fileparts(ROINii{rois-1});
end
ROIstat.ROIname{rois} = name;
for par = 1 : length(IVIMPars)
eval(['ROIstat.' IVIMPars{par} '.mean(rois) = nanmean(' IVIMPars{par} '(ROIs{rois}==1),''all'');']);
eval(['ROIstat.' IVIMPars{par} '.median(rois) = nanmedian(' IVIMPars{par} '(ROIs{rois}==1),''all'');']);
eval(['ROIstat.' IVIMPars{par} '.std(rois) = nanstd(' IVIMPars{par} '(ROIs{rois}==1),0,''all'');']);
eval(['ROIstat.' IVIMPars{par} '.CV(rois) = ROIstat.' IVIMPars{par} '.std(rois) /ROIstat.' IVIMPars{par} '.mean(rois);']);
eval(['ROIstat.' IVIMPars{par} '.iqr(rois) = iqr(reshape(' IVIMPars{par} '(ROIs{rois}==1),[],1));']);
eval(['ROIstat.' IVIMPars{par} '.q1(rois) = prctile(reshape(' IVIMPars{par} '(ROIs{rois}==1),[],1),25);']);
eval(['ROIstat.' IVIMPars{par} '.q3(rois) = prctile(reshape(' IVIMPars{par} '(ROIs{rois}==1),[],1),1);']);
end
end
filenm = sprintf([P.outputFolder '%sIDEALfit%s_steps_%s.mat'],filesep, file_name, num2str(size(P.Dims_steps, 1)));
save(filenm);
end