-
Notifications
You must be signed in to change notification settings - Fork 0
/
index.html
4164 lines (3610 loc) · 318 KB
/
index.html
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
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
<!DOCTYPE HTML>
<!--
Editorial by Pixelarity
pixelarity.com | hello@pixelarity.com
License: pixelarity.com/license
-->
<html lang="en">
<head>
<title>CAT12 Manual</title>
<meta charset="utf-8">
<link rel="stylesheet" href="css/main.css" >
<!-- ############################# -->
</head>
<body class="is-preload">
<!-- Wrapper -->
<div id="wrapper">
<!-- Main -->
<div id="main">
<div class="inner">
<!-- Content -->
<section>
<!-- Introduction -->
<div id="content">
<h1>Manual Computational Anatomy Toolbox CAT12</h1>
<!-- ################################################################################# -->
<h2 id="intro">Introduction and overview</h2>
<!-- ############################# -->
<h4 id="intro_cat12">Introduction</h4>
<p class="image right"><img width=100% alt="Human Brain" src="images/human_brain.png"></p>
<p>
The brain is the most complex organ of the human body, and no two brains are alike. The study of the human brain is still in its infancy, but rapid technical advances in image acquisition and image processing have allowed for ever more refined characterizations of its micro- and macro-structure. Enormous efforts, for example, have been made to map differences between groups (e.g., young vs. old, diseased vs. healthy, male vs. female), to capture changes over time (e.g., from infancy to old age, or in the framework of neuroplasticity, as a result of a clinical intervention), or to assess correlations of brain attributes (e.g., measures of length, volume, shape) with behavioral, cognitive, or clinical parameters.
</p>
<div class="image left"><img width=60% alt="Logo" src="images/cat_logo.png"></div>
<p>
CAT12 is a powerful suite of tools for morphometric analyses with an intuitive graphical user interface, but also usable as a shell script. CAT is suitable for beginners, casual users, experts, and developers alike providing a comprehensive set of analysis options, workflows, and integrated pipelines. The available analysis streams allow for voxel-based, surface-based, as well as region-based morphometric analyses. Importantly, CAT includes various quality control options and covers the entire analysis workflow, from cross-sectional or longitudinal data processing, to the statistical analysis, and visualization of results.
</p>
<!-- ############################# -->
<h4 id="overview">Overview about the manual</h4>
<p>This manual is intended to help any user to perform a computational anatomy analysis using the CAT12 Toolbox. Although it mainly focuses on voxel-based morphometry (VBM) other variants of computational analysis such as deformation-based morphometry (DBM), surface-based morphometry (SBM), and region of interest (ROI) morphometric analysis are also presented and can be applied with few changes.</p>
The manual can be divided into four main sections:<br>
<ul>
<li>Naturally, a quick guide of how to <a href="#get_started">get started</a> is given at the beginning. This section provides information about <a href="#download">downloading and installing</a> the software and <a href="#starting">starting</a> the Toolbox. Also, a <a href="#basic_vbm">brief overview</a> of the steps of a VBM analysis is given.</li>
<li>This is followed by a <a href="#process_data">detailed description of a basic VBM analysis</a> that guides the user step-by-step through the entire process - from preprocessing to contrast selection. This description should provide all the information necessary to successfully analyze most studies.</li>
<li>There are some specific cases of VBM analyses, for which the basic analysis workflow has to be adapted. These cases are <a href="#long">longitudinal studies</a> and studies in <a href="#children">children or special patient populations</a>. Relevant changes to a basic VBM analysis are described here and how these changes can be applied. Only the changes are described here - steps such as quality control or smoothing are the same as those described in the basic analysis and are not repeated a second time.</li>
<li>The guide concludes with <a href="#add_info">additional information</a> about spaces after registration, naming conventions used and other hints.</li>
</ul>
<!-- ################################################################################# -->
<h2 id="quick">Quick start guide</h2>
<!-- ############################# -->
<h4 id="vbm_data">VBM data</h4>
<ul>
<li><a class="function" href="matlab:spm_jobman('interactive','','spm.tools.cat.estwrite');">Segment</a> data using defaults (use <a class="function" href="matlab:spm_jobman('interactive','','spm.tools.cat.long');">Segment Longitudinal Data</a> for longitudinal data).<br>
The resulting segmentations that can now be used for VBM are saved in the "mri" folder and are named "mwp1" for gray matter and "mwp2" for white matter. If you have used the longitudinal pipeline, the default segmentations for gray matter are named "mwp1r" or "mwmwp1r" if the longitudinal model for detecting larger changes was selected.</li>
<li>Get total intracranial volume<a class="function" href="matlab:spm_jobman('interactive','','spm.tools.cat.tools.calcvol');"> (TIV)</a> to correct for different brain sizes and volumes.<br>
Select the xml-files that are saved in the "report" folder.</li>
<li>Check the data quality with <a class="function" href="matlab:spm_jobman('interactive','','spm.tools.cat.tools.check_homogeneity');">Sample Homogeneity</a> for VBM data (optionally consider TIV and age as nuisance variables).<br>
Select the gray or white matter segmentations from the first step.</li>
<li><a class="function" href="matlab:spm_jobman('interactive','','spm.spatial.smooth');">Smooth</a> data (recommended start value 6mm<sup>1</sup>).<br>
Select the gray or white matter segmentations from the first step.</li>
<li>Specify the <a class="function" href="matlab:spm_jobman('interactive','','spm.tools.cat.factorial_design');">Basic Models</a> with the smoothed gray or white matter segmentations and check for design orthogonality and sample homogeneity:
<ul>
<li>Select "Two-sample t-test" or "Multiple regression" or use "Full factorial" for any cross-sectional data.</li>
<li>Use "Flexible factorial" for longitudinal data.</li>
<li>Use TIV as covariate (confound) to correct different brain sizes.</li>
<li>Select threshold masking with an absolute value of 0.1. This threshold can ultimately be increased to 0.2 or even 0.25 if you still notice non-brain areas in your analysis.</li>
<li>If you find a considerable correlation between TIV and any other parameter of interest it is advisable to use global scaling with TIV. For more information, refer to <a href="index.html#ortho">Orthogonality</a>.</li>
</ul></li>
<li><a class="function" href="matlab:spm_jobman('interactive','','spm.stats.fmri_est');">Estimate</a> the model and finally call <a class="function" href="matlab:[hReg xSPM SPM] = cat_spm_results_ui('Setup');">Results</a>.</li>
<li>Optionally <a class="function" href="matlab:spm_jobman('interactive','','spm.tools.cat.tools.T2x');">Transform and Threshold SPM-maps</a> to (log-scaled) p-maps or correlation maps.</li>
<li>Optionally, you can try <a class="function" href="matlab:if exist('spm_TFCE'),spm_jobman('interactive','','spm.tools.tfce_estimate'), else cat_install_tfce, end;">Threshold-Free Cluster Enhancement</a> (TFCE) with the SPM.mat file of a previously estimated statistical design.</li>
<li>Optionally <a class="function" href="matlab:cat_vol_slice_overlay;">Overlay Selected Slices</a>. If you are using log-p scaled maps from "Transform SPM-maps" without thresholds or the TFCE_log maps, use the following values as the lower range for the colormap for the thresholding: 1.3 (P<0.05); 2 (P<0.01); 3 (P<0.001).</li>
<li>Optionally use <a class="function" href="matlab:y=cat_surf_results;">Surface Overlay</a> for visualization of your results. Select the results (preferably saved as log-p maps with "Transform SPM-maps" or the TFCE_log maps with the different methods for multiple comparison correction) to display rendering views, slice overlay, and a glassbrain of your results.</li>
<li>Optionally estimate results for ROI analysis using <a class="function" href="matlab:cat_stat_analyze_ROIs;">Analyze ROIs</a>. Here, the SPM.mat file of the already estimated statistical design will be used. For more information, see <a href="#atlas">Atlas Creation and ROI based Analysis</a>.</li>
</ul>
<!-- ############################# -->
<h4 id="add_surf">Additional surface data</h4>
<ul>
<li><a class="function" href="matlab:spm_jobman('interactive','','spm.tools.cat.estwrite');">Segment</a> data and also select "Surface and thickness estimation" under "Writing options" (for longitudinal data use <a class="function" href="matlab:spm_jobman('interactive','','spm.tools.cat.long');">Segment Longitudinal Data</a>).<br>
The surfaces data are saved in the folder "surf" and are named "?h.thickness.*" for cortical thickness.</li>
<li>Optionally, <a class="function" href="matlab:spm_jobman('interactive','','spm.tools.cat.stools.surfextract');">Extract Additional Surface Parameters</a> (e.g. sulcal depth, gyrification, cortical complexity).</li>
<li><a class="function" href="matlab:spm_jobman('interactive','','spm.tools.cat.stools.surfresamp');">Resample & smooth</a> surface data (suggested starting value 12mm for cortical thickness and sulcal depth and 20-25mm for folding measures<b><sup>1</sup></b>, use the default merging of hemispheres).<br>
Select the "lh.thickness.*" data in the folder "surf". The resampled data are named "s12.mesh.resampled_32k.thickness.*" for 12mm smoothed, merged hemispheres that were resampled to 32k template space.</li>
<li>Check data quality of the resampled data using <a class="function" href="matlab:spm_jobman('interactive','','spm.tools.cat.tools.check_homogeneity');">Sample Homogeneity</a>.</li>
<li>Build <a class="function" href="matlab:spm_jobman('interactive','','spm.tools.cat.factorial_design');">Basic Models</a> for the resampled data and check for design orthogonality and sample homogeneity.
<ul>
<li>Select "Two-sample t-test" or "Multiple regression" or use "Full factorial" for any cross-sectional data.</li>
<li>Use "Flexible factorial" for longitudinal data.</li>
<li>It is not necessary to use TIV as a covariate (confound) because cortical thickness or other surface values are usually not dependent on TIV.</li>
<li>It is not necessary to use any threshold masking.</li>
<li>If you find a considerable correlation between a nuisance parameter and any other parameter of interest it is advisable to use global scaling with that parameter. For more information, refer to <a href="index.html#ortho">Orthogonality</a>.</li>
</ul></li>
<li><a class="function" href="matlab:cat_stat_spm;">Estimate</a> the surface model and finally call <a class="function" href="matlab:[hReg xSPM SPM] = cat_spm_results_ui('Setup');">Results</a>.</li>
<li>Optionally <a class="function" href="matlab:spm_jobman('interactive','','spm.tools.cat.tools.T2x_surf');">Transform and Threshold SPM-maps</a> to (log-scaled) p-maps or correlation maps.</li>
<li>Optionally, you can try <a class="function" href="matlab:if exist('spm_TFCE'),spm_jobman('interactive','','spm.tools.tfce_estimate'), else cat_install_tfce, end;">Threshold-Free Cluster Enhancement</a> (TFCE) with the SPM.mat file of a previously estimated statistical design.</li>
<li>Optionally use <a class="function" href="matlab:y=cat_surf_results;">Surface Overlay</a> for visualization of your results. Select the results (preferably saved as log-p maps with "Transform SPM-maps" or the TFCE_log maps with the different methods for multiple comparison correction) for the merged hemispheres to display rendering views of your results.</li>
<li>Optionally <a class="function" href="matlab:spm_jobman('interactive','','spm.tools.cat.stools.surf2roi');">Extract ROI-based Surface Values</a> such as gyrification or fractal dimension to provide ROI analysis. Since version 12.7 extraction of ROI-based thickness is not necessary anymore because this is now included in the segmentation pipeline.</li>
<li>Optionally estimate results for ROI analysis using <a class="function" href="matlab:cat_stat_analyze_ROIs;">Analyze ROIs</a>. Here, the SPM.mat file of the already estimated statistical design will be used. For more information, see <a href="#atlas">Atlas Creation and ROI based Analysis</a>.</li>
</ul>
<!-- ############################# -->
<h5 id="add_opts">Additional options</h5>
<p>Additional parameters and options are displayed in the <a class="function" href="matlab:cat12('expert');">CAT12 expert mode</a>. Please note that this mode is for experienced users only.</p>
<!-- ############################# -->
<h5 id="errors">Errors during preprocessing</h5>
<p>Please use the <a class="function" href="matlab:cat_io_senderrormail;">Report Error</a> function if any errors during preprocessing occurred. You first have to select the "err" directory, which is located in the folder of the failed record, and finally, the specified zip-file should be attached manually in the mail.</p>
<!-- ############################# -->
<h5><sup>1</sup>Note to filter sizes for Gaussian smoothing</h5>
<p>Smoothing increases the signal-to-noise ratio. The matched filter theorem states that the filter that matches your signal maximizes the signal-to-noise ratio. If we expect signal with a Gaussian shape and a FWHM of, say 6mm in our images, then this signal will be best detected after we smooth our images with a 6mm FWHM Gaussian filter. If you are trying to study small areas like the amygdala, we can use smaller filter sizes because of the small size of these regions. For larger regions like the cerebellum, the effects are better detected with larger filters.<br>
Please also note that for the analysis of cortical folding measures such as gyrification or cortical complexity the filter sizes have to be larger (i.e. in the range of 15-25mm). This is due to the underlying nature of this measure that reflects contributions from both sulci as well as gyri. Therefore, the filter size should exceed the distance between a gyral crown and a sulcal fundus.</p>
<!-- ################################################################################# -->
<h2 id="version">Version information</h2>
<p>
Preprocessing should remain unaffected until the next minor version number (12.x). New processing of your data is not necessary if the minor version number of CAT12 remains unchanged.
</p>
<!-- ###################################### -->
<h3>Changes in version CAT12.9 (2550)</h3>
<details>
<summary><u>Changes in preprocessing pipeline (which affects your results compared to CAT12.8.2)</u></summary>
<ul>
<li>The cortical surface extraction has been extensively updated with many changes that result in more reliable thickness measurements and better handling of partial volume effects.</li>
<li>Initial affine registration now uses a brain mask to obtain more reliable estimates, resulting in more accurate non-linear spatial registration.</li>
<li>Various smaller changes have been made for blood vessel detection and background detection for skull-stripping.</li>
<li><a class="function" href="matlab:spm_jobman('interactive','','spm.tools.cat.long');">Segment Longitudinal Data</a> is now saving both plasticity and aging models by default.</li>
<li>The previous cat_vol_sanlm from r1980 has been rescued and renamed to cat_vol_sanlm2180.m, which performs significantly better.</li>
<li>The quartic mean Z-score is now used to assess sample homogeneity as it gives greater weight to outliers and makes them easier to identify in the plot.</li>
<li>Several changes have been made to support Octave compatibility.</li>
</ul>
</details>
<!-- ###################################### -->
<h3>Changes in version CAT12.8.2 (2130)</h3>
<details>
<summary><u>Important new features</u></summary>
<ul>
<li>The CAT12 manual is now converted to HTML and merged with the online help to a single HTML file with interactive Matlab links that can be called in Matlab.</li>
<li>Added new thalamic nuclei <a class="function" href="matlab:cat_vol_display_label(struct('data',{{fullfile(fileparts(which('cat12')),'templates_MNI152NLin2009cAsym','Template_T1.nii'); fullfile(fileparts(which('cat12')),'templates_MNI152NLin2009cAsym','thalamic_nuclei.nii')}}))">atlas</a>.</li>
<li>Moved some rarely used CAT12 options to expert mode.</li>
<li>Added many new options to <a class="function" href="matlab:y=cat_surf_results;">visualize 3D (VBM) results</a> (e.g. new glass brain).</li>
<li>The tool to check <a class="function" href="matlab:spm_jobman('interactive','','spm.tools.cat.tools.check_homogeneity');">Sample Homogeneity</a> has many new options and is now based on calculation of Z-score which is much faster for larger samples.</li>
<li>New <a class="function" href="matlab:spm_jobman('interactive','','spm.tools.cat.factorial_design');">Basic Models</a> to easier define cross-sectional and longitudinal statistical designs. Some unnecessary options have been removed compared to "Basic Models" in SPM12 and some useful options have been added.</li>
</ul>
</details>
<!-- ###################################### -->
<h3>Changes in version CAT12.8.1 (1975)</h3>
<details>
<summary><u>Changes in preprocessing pipeline (which affects your results compared to CAT12.8)</u></summary>
<ul>
<li>The longitudinal pipeline has been largely updated and is no longer compatible with preprocessing with version 12.8:
<ul>
<li>The estimate of subject-specific TPM should now be more stable and less sensitive to changes between time points.</li>
<li>For data where the brain/head is still growing between time points and major changes are expected, a new longitudinal development model was added. For this type of data, an adapted pipeline was created that is very similar to the longitudinal aging model, but uses a time point-independent affine registration to adjust for brain/head growth. In addition, this model uses a subject-specific TPM based on the average image.</li>
<li>An additional longitudinal report is now provided to better assess differences between time points.</li>
</ul></li>
</ul>
</details>
<!-- ###################################### -->
<h3>Changes in version CAT12.8 (1830)</h3>
<details>
<summary><u>Changes in preprocessing pipeline (which affects your results compared to CAT12.7)</u></summary>
<ul>
<li>Volumetric templates, atlases, and TPMs are now transformed to MNI152NLin2009cAsym space to better match existing standards. The templates_volume folder is now renamed to ''templates_MNI152NLin2009cAsym'' to indicate the template space used. The Dartel and Geodesic Shooting templates are renamed or relocated:
<ul>
<li>templates_volumes/Template_0_IXI555_MNI152_GS.nii → templates_MNI152NLin2009cAsym/Template_0_GS.nii</li>
<li>templates_volumes/Template_1_IXI555_MNI152.nii → templates_MNI152NLin2009cAsym/Template_1_Dartel.nii</li>
<li>templates_volumes/TPM_Age11.5.nii → templates_MNI152NLin2009cAsym/TPM_Age11.5.nii</li>
<li>templates_volumes/Template_T1_IXI555_MNI152_GS.nii → templates_MNI152NLin2009cAsym/Template_T1.nii</li>
<li>spm12/toolbox/FieldMap/T1.nii → templates_MNI152NLin2009cAsym/T1.nii</li>
<li>spm12/toolbox/FieldMap/brainmask.nii → templates_MNI152NLin2009cAsym/brainmask.nii</li>
</ul></li>
<li>The volumetric atlases have been revised and are now defined with a spatial resolution of 1mm, except for the Cobra atlas, which is defined with 0.6mm resolution. The labels of the original atlases were either transformed from the original data or recreated using a maximum likelihood approach when manual labels were available for all subjects (Cobra, LPBA40, IBSR, Hammers, Neuromorphometrics). In addition, the original labels are now used for all atlases if possible. Some atlases were updated to include new regions (Julichbrain, Hammers) and a new atlas of thalamic nuclei was added. Please note that this will also result in slight differences in ROI estimates compared to previous versions.</li>
<li>The bounding box of the Dartel and Geodesic Shooting templates has been changed, resulting in a slightly different image size of the spatially registered images (i.e. modulated normalized segmentations). Therefore, older preprocessed data should not (and cannot) be mixed with the new processed data (which is intended).</li>
<li>Transformed T1 Dartel/GS surface templates to the new MNI152NLin2009cAsym space:
<ul>
<li>templates_surfaces/lh.central.Template_T1_IXI555_MNI152_GS.gii → templates_surfaces/lh.central.Template_T1.gii </li>
<li>templates_surfaces/rh.central.Template_T1_IXI555_MNI152_GS.gii → templates_surfaces/rh.central.Template_T1.gii </li>
<li>templates_surfaces_32k/lh.central.Template_T1_IXI555_MNI152_GS.gii → templates_surfaces_32k/lh.central.Template_T1.gii </li>
<li>templates_surfaces_32k/rh.central.Template_T1_IXI555_MNI152_GS.gii → templates_surfaces_32k/rh.central.Template_T1.gii </li>
</ul></li>
<li>The surface pipeline has been optimized to better handle data at different spatial resolutions.</li>
<li>Older preprocessing pipelines (12.1, 12.3, 12.6) were removed because their support became too difficult.</li>
</ul>
</details>
<details>
<summary><u>Important new features</u></summary>
<ul>
<li>The Mahalanobis distance in the <a class="function" href="matlab:spm_jobman('interactive','','spm.tools.cat.tools.check_homogeneity');">Quality Check</a> is now replaced by the normalized ratio between overall weighted image quality (IQR) and mean correlation. A low ratio indicates good quality before and after preprocessing and means that IQR is highly rated (resulting in a low nominal number/grade) and/or mean correlation is high. This is hopefully a more intuitive measure to combine image quality measurement before and after preprocessing.</li>
<li>CAT12 now allows the use of the BIDS directory structure for storing data (not possible for the longitudinal pipeline). A BIDS path can be defined relative to the participant level directory. The segmentation module now supports the input of nii.gz files (not possible for the longitudinal pipeline).</li>
<li>The <a class="function" href="matlab:spm_jobman('interactive','','spm.tools.cat.factorial_design');">Basic Models</a> function has been completely restructured and simplified. There are now only two models available for: (1) cross-sectional data and (2) longitudinal data. Options that are not relevant for VBM or SBM have been removed. In addition, a new experimental option has been added that allows a voxel-wise covariate to be defined. This can be used (depending on the contrast defined) to (1) remove the nuisance effect of structural data (e.g. GM) on functional data or (2) examine the relationship (regression) between functional and structural data. Additionally, an interaction can be modeled to investigate whether the regression between functional and structural data differs between two groups. Please note that the saved vSPM.mat file can only be evaluated with the TFCE toolbox.</li>
<li>Added a new function cat_io_data2mat.m to save spatially registered volume or resampled surface data as Matlab data matrix for further use with machine learning tools. Volume data can be resampled to lower spatial resolutions and can optionally be masked to remove non-brain areas.</li>
<li>Added a new function <a class="function" href="matlab:spm_jobman('interactive','','spm.tools.cat.tools.ROIsum');">cat_vol_ROI_summarize.m</a> to summarise co-registered volume data within a region of interest (ROI). This tool can be used in order to estimate ROI information for other (co-registered) modalities (i.e. DTI, (rs)fMRI) which can be also defined as 4D data. Several predefined summary functions are available, as well as the possibility to define your own function.</li>
<li>Added a new function <a class="function" href="matlab:spm_jobman('interactive','','spm.tools.cat.tools.quality_measures');">cat_stat_quality_measures.m</a> to estimate and save quality measures for very large samples.</li>
<li>Added standalone tools for de-facing, DICOM import, and estimating and saving quality measures for large samples.</li>
</ul>
</details>
<!-- ###################################### -->
<h3>Changes in version CAT12.7 (1700)</h3>
<details>
<summary><u>Changes in preprocessing pipeline (which affects your results compared to CAT12.6)</u></summary>
<ul>
<li>Geodesic shooting registration and surface estimation are now used by default.</li>
<li>The surface pipeline is largely updated: (1) Parameters for surface reconstruction were optimized. (2) Freesurfer distance method Tfs is now implemented, which is computed as the average of the closest distances from the pial to the white matter surface and from the white matter to the pial surface. This reduces the occurrence of larger thickness values and results in more reliable thickness measures. For mapping of 3D data the old thickness metric from PBT is more appropriate and is still used.</li>
<li>An additional longitudinal model is implemented that also takes into account deformations between time points. The use of deformations between the time points makes it possible to estimate and detect larger changes, while subtle effects over shorter periods of time in the range of weeks or a few months can be better detected with the model for small changes.</li>
<li>Minor changes were made to the segmentation approach to improve accuracy and reliability.</li>
<li>Internal resampling is now using a new ''optimal'' resolution setting to better support high-resolution data.</li>
<li>Changed recommendation and defaults for smoothing size to smaller values.</li>
<li>Renamed template folder for volumes to templates_volumes.</li>
<li>Atlases installed in spm12/atlas are now called cat12_atlasname.</li>
<li>The old AAL atlas has been replaced by the AAL3 atlas.</li>
<li>Cobra atlas is updated because of some previous inconsistencies.</li>
</ul>
</details>
<details>
<summary><u>Important new features</u></summary>
<ul>
<li>New GUI</li>
<li>Added older preprocessing pipelines (12.1, 12.3, 12.6) to provide compatible versions to previous preprocessing. These older preprocessing pipelines are available via the <a class="function" href="matlab:spm_jobman('interactive')">SPM batch editor</a> (SPM → Tools → CAT12) or through expert mode.</li>
<li>Added simple batch for cross-sectional and longitudinal data that combines several processing steps from preprocessing to smoothing. These batches are available via the <a class="function" href="matlab:spm_jobman('interactive')">SPM batch editor</a>(SPM → Tools → CAT12) or through expert mode.</li>
<li>The function <a class="function" href="matlab:y=cat_surf_results;">Display surface results</a> can now also visualize results from VBM analysis and map the 3D volume information onto the surface using an absmax mapping function inside the cortical band. The function is now renamed to ''Surface Overlay''.</li>
<li>You can call <a class="function" href="matlab:[hReg xSPM SPM] = cat_spm_results_ui('Setup');">Results</a> now from the CAT12 GUI with some new functionality for surfaces and the option to call TFCE results.</li>
<li>CAT12 now uses Piwik for anonymized CAT12 user statistics (i.e. version information, potential errors). See <a href="cat_statistics.html">CAT12 user statistics</a> for more information. This can be optionally disabled in <a class="function" href="matlab:edit('cat_defaults');">cat_defaults.m</a>.</li>
<li>The extraction of ROI-based thickness is not necessary anymore because this is now included in the segmentation pipeline in cat_main.m.</li>
<li>Changed gifti-format after resampling to use external dat-files. This increases processing speed and prevents the SPM.mat file for surfaces from becoming too large. This can be optionally disabled in <a class="function" href="matlab:edit('cat_defaults');">cat_defaults.m</a>.</li>
<li>The use of own atlases for ROI processing is now supported.</li>
<li>Updated and extended examples for statistical designs and respective contrasts in the CAT12 manual.</li>
</ul>
</details>
<!-- ###################################### -->
<h3>Changes in version CAT12.6 (1445)</h3>
<details>
<summary><u>Changes in preprocessing pipeline (which affects your results compared to CAT12.5)</u></summary>
<ul>
<li>Two main parts of the preprocessing of CAT12 were largely updated: (1) Incorrect estimates of the initial affine registration were found to be critical for all subsequent preprocessing steps and mainly concerned skull-stripping and tissue segmentation. This was a particular problem in the brains of older people or children, where the thickness of the skull differs from that of the template. The new estimate of the initial affine registration should now be more robust. In the CAT report, the registered contour of the skull and the brain is now overlayed onto the image to allow for easier quality control. (2) Skull-stripping now uses a new adaptive probability region-growing (APRG) approach, which should also be more robust. APRG refines the probability maps of the SPM approach by region-growing techniques of the gcut approach with a final surface-based optimization strategy. This is currently the method with the most accurate and reliable results.</li>
<li>The longitudinal pipeline should now also be more sensitive for the detection of effects over longer time periods with VBM (ROI and SBM approaches are not affected by the length of the period). In earlier versions, the average image was used to estimate the spatial registration parameters for all time points. Sometimes this average image was not as accurate if the images of a subject were too different (e.g. due to large ventricular changes). Now, we rather use the average spatial registration parameters (i.e. deformations) of all time points, which makes the approach more robust for longer periods of time. However, the SPM12 Longitudinal Toolbox can be a good alternative for longer periods of time if you want to analyze your data voxel by voxel. Surface-based preprocessing and also the ROI estimates in CAT12 are not affected by the potentially lower sensitivity to larger changes, as the realigned images are used independently to create cortical surfaces, thickness, or ROI estimates.</li>
</ul>
</details>
<details>
<summary><u>Important new features</u></summary>
<ul>
<li>CAT report now additionally plots the contour of the registered skull and brain onto the image and visualizes skull-stripping. <a class="function" href="matlab:y=cat_surf_results;">Display surface results</a> is largely updated. </li>
<li>Parallelization options in CAT12 now enable subsequent batch jobs and are also supported for longitudinal preprocessing.</li>
</ul>
</details>
<!-- ###################################### -->
<h3>Changes in version CAT12.5 (1355)</h3>
<details>
<summary><u>Changes in preprocessing pipeline (which affects your results compared to CAT12.3)</u></summary>
<ul>
<li>Detection of white matter hyperintensities (WMHs) is updated and again enabled by default.</li>
<li>The default internal interpolation setting is now "Fixed 1 mm" and offers a good trade-off between optimal quality and preprocessing time and memory demands. Standard structural data with a voxel resolution around 1 mm or even data with high in-plane resolution and large slice thickness (e.g. 0.5x0.5x1.5 mm) will benefit from this setting. If you have higher native resolutions the highres option "Fixed 0.8 mm" will sometimes offer slightly better preprocessing quality with an increase of preprocessing time and memory demands.</li>
</ul>
</details>
<details>
<summary><u>Important new features</u></summary>
<ul>
<li>CAT12 can now deal with lesions that have to be set to zero in your image using the Stroke Lesion Correction (SLC) in expert mode. These lesion areas are not used for segmentation or spatial registration, thus these preprocessing steps should be almost unaffected.</li>
</ul>
</details>
<!-- ###################################### -->
<h3>Changes in version CAT12.4 (1342)</h3>
<ul>
<li>This version had some severe errors in spatial registration which affected all spatially registered data and should not be used anymore.</li>
</ul>
<!-- ###################################### -->
<h3>Changes in version CAT12.3 (1310)</h3>
<details>
<summary><u>Changes in preprocessing pipeline (which affects your results compared to CAT12.2)</u></summary>
<ul>
<li>Skull-stripping is again slightly changed and the SPM approach is now used as default. The SPM approach works quite well for the majority of data. However, in some rare cases, parts of GM (i.e. in the frontal lobe) might be cut. If this happens the GCUT approach is a good alternative.</li>
<li>Spatial adaptive non-local mean (SANLM) filter is again called as a very first step because noise estimation and de-noising works best for original (non-interpolated) data.</li>
<li>Detection of white matter hyperintensities (WMHs) is currently disabled by default, because of unreliable results for some data.</li>
</ul>
</details>
<details>
<summary><u>Important new features</u></summary>
<ul>
<li>Cobra atlas has been largely extended and updated.</li>
</ul>
</details>
<!-- ###################################### -->
<h3>Changes in version CAT12.2 (1290)</h3>
<details>
<summary><u>Changes in preprocessing pipeline (which affects your results compared to CAT12.1)</u></summary>
<ul>
<li>Skull-stripping now additionally uses SPM12 segmentations by default: The default gcut approach in CAT12.1 removed too much of the surrounding (extracranial) CSF, which led to a slight underestimation of TIV for atrophied brains. The skull-stripping approach based on the SPM12 segmentations prevents this through a more conservative approach. However, sometimes parts of the meninges (i.e. dura mater) or other non-brain parts remain in the GM segmentation. By combining both approaches a more reliable skull-stripping is achieved.</li>
<li>More reliable estimation of TIV: The changed skull-stripping also affects the estimation of TIV, which is now more reliable, especially for atrophied brains.</li>
</ul>
</details>
<details>
<summary><u>Important new features</u></summary>
<ul>
<li>Automatic check for design orthogonality and sample homogeneity using SPM.mat in <a class="function" href="matlab:spm_jobman('interactive','','spm.tools.cat.factorial_design');">Basic Models</a></li>
<li>Added equi-volume model by Bok and a multi-save option for <a class="function" href="matlab:spm_jobman('interactive','','spm.tools.cat.stools.vol2surf');">mapping native volumes to individual surfaces.</a></li>
<li>Added <a href="https://github.com/ThomasYeoLab/CBIG/tree/master/stable_projects/brain_parcellation/Schaefer2018_LocalGlobal">Local-Global Intrinsic Functional Connectivity parcellation</a> by Schaefer et al. for resting-state fMRI data.</li>
</ul>
</details>
<!-- ################################################################################# -->
<h2 id="get_started">Getting started</h2>
<!-- ############################# -->
<h4 id="download">Download and installation</h4>
<ul>
<li>The CAT12 Toolbox runs within SPM12. That is, SPM12 must be installed and added to your Matlab search path before the CAT12 Toolbox can be installed (see the <a href="http://www.fil.ion.ucl.ac.uk/spm/">SPM site</a> and <a href="http://en.wikibooks.org/wiki/SPM">SPM Wiki</a> for more help).
</li>
<li><a href="https://neuro-jena.github.io/cat/index.html#DOWNLOAD">Download</a> and unzip the CAT12 Toolbox. You will get a folder named "cat12", which contains various Matlab files and compiled scripts. Copy the folder "cat12" into the SPM12 "toolbox" folder.
</li>
<li>On newer MAC OS systems (10.15 or higher: Catalina, Big Sur, Monterey, Ventura) you may need to follow the <a href="https://en.wikibooks.org/wiki/SPM/Installation_on_64bit_Mac_OS_(Intel)#macOS_Catalina">SPM12 installation</a> instructions to prevent issues with MEX files.
</li>
<li>Older CAT12 versions can be found <a href="http://www.neuro.uni-jena.de/cat12/">here</a>.</li>
</ul>
<!-- ############################# -->
<h4 id="starting">Starting the toolbox</h4>
<ul>
<li>Start Matlab</li>
<li>Start SPM12 (i.e., type "spm pet")</li>
<li>Select "cat12" from the SPM menu (see <a href="#figure1">Figure 1</a>). You will find the drop-down menu between the "Display" and the "Help" button (you can also call the Toolbox directly by typing "cat12" on the Matlab command line). This will open the CAT12 Toolbox as an additional window (Fig. 2).</li>
</ul>
<table class="img">
<tr>
<td colspan="1" rowspan="1">
<p><img id="figure1" width=70% alt="spm_gui" src="images/spm_gui.png"></p>
</td>
<td colspan="1" rowspan="1">
<p><img id="figure2" width=90% alt="cat_gui" src="images/cat_gui.png"></p>
</td>
</tr>
<tr>
<td colspan="1" rowspan="1">
<figcaption><b>Figure 1:</b> SPM12 menu</figcaption>
</td>
<td colspan="1" rowspan="1">
<figcaption><b>Figure 2:</b> CAT12 Window</figcaption>
</td>
</tr>
</table>
<!-- ############################# -->
<h4 id="basic_vbm">Basic VBM analysis (overview)</h4>
<p>The CAT12 Toolbox comes with different modules, which may be used for analysis. Usually, a VBM analysis comprises the following steps:</p>
<u>(a) Preprocessing:</u><br><br>
<ol>
<li>T1 images are <b>normalized</b> to a template space and <b>segmented</b> into gray matter (GM), white matter (WM), and cerebrospinal fluid (CSF). The preprocessing parameters can be adjusted via the module "Segment Data".</li>
<li>After the preprocessing is finished, a <b>quality check</b> is highly recommended. This can be achieved via the modules "Display slices" and "Check sample". Both options are located in the CAT12 window under "Check Data Quality". Furthermore, quality parameters are estimated and saved in xml-files for each data set during preprocessing. These quality parameters are also printed on the report PDF-page and can be additionally used in the module "Check sample".</li>
<li>Before entering the GM images into a statistical model, image data needs to be <b>smoothed</b>. Of note, this step is not implemented into the CAT12 Toolbox but achieved via the standard SPM module "Smooth".</li>
</ol>
<u>(b) Statistical analysis:</u><br><br>
<ol>
<li>The smoothed GM images are entered into a statistical analysis. This requires building a statistical model (e.g. T-Tests, ANOVAs, multiple regressions). This is done by the standard SPM modules "Specify 2nd Level" or preferably "Basic Models" in the CAT12 window covering the same function but providing additional options and a simpler interface optimized for structural data.</li>
<li>The statistical model is estimated. This is done with the standard SPM module "Estimate" (except for surface-based data where the function "Estimate Surface Models" should be used instead).</li>
<li>If you have used total intracranial volume (TIV) as a confound in your model to correct for different brain sizes it is necessary to check whether TIV reveals a considerable correlation with any other parameter of interest and rather uses global scaling as an alternative approach.</li>
<li>After estimating the statistical model, contrasts are defined to get the results of the analysis. This is done with the standard SPM module "Results".</li>
</ol>
<p><em>The sequence of "preprocessing → quality check → smoothing → statistical analysis" remains the same for every VBM or SBM analysis, even when different steps are adapted (see <a href="#children">Processing of Data such as Children</a>).</em></p>
<b>A few words about the Batch Editor...</b><br>
<ul>
<li>As soon as you select a module from the CAT12 Toolbox menu, a new window (the Batch Editor) will open. The Batch Editor is the environment where you will set up your analysis (see <a href="#figure3">Figure 3</a>). For example, an "<-X" indicates where you need to select files (e.g. your image files, the template, etc.). Other parameters have either default settings (which can be modified) or require input (e.g. choosing between different options, providing text or numeric values, etc.).</li>
<li>Once all missing parameters are set, a green arrow will appear on the top of the window (the current snapshots in <a href="#figure3">Figure 3</a> show the arrow still in gray). Click this arrow to run the module or select "File → Run Batch". It is very useful to save the settings before you run the batch (click on the disk symbol or select "File → Save Batch").
<div class="box right"><sup>*</sup> Additional CAT12-related information can be found by selecting "VBM Website" in the CAT12 window (Tools → Internet → VBM Website). This will open a website. Here, look for "VBM subpages" on the right.</div></li>
<li>Of note, you can always find helpful information and parameter-specific explanations at the bottom of the Batch Editor window<b><sup>*</sup></b>.
</li>
<li>All settings can be saved either as .mat file or as .m script file and reloaded for later use. The .m script file has the advantage of being editable with a text editor.</li>
</ul>
<br><p><img id="figure3" width=80% alt="Batch Editor" src="images/batch_gui.jpg"></p>
<figcaption><b>Figure 3:</b> The Batch Editor is the environment where the analysis is set up. Left: For all settings marked with "<-X", files have to be selected ("Select Files"). Right: Parameters can be edited and adapted ("Edit Value").</figcaption>
<!-- ################################################################################# -->
<h2 id="overview_process">Overview of CAT12 processing</h2>
<!-- ############################# -->
<h4 id="major_process">CAT12 major processing steps</h4>
<p>As shown in <a href="#figure4">Figure 4</a>, CAT's processing workflow comprises two main steps: voxel-based processing and surface-based processing. The former is a prerequisite for the latter, but not the other way round. That is, while voxel-based processing is always required for surface-based analyses, users not interested in surface-based analyses can simply omit this second step to save processing time. The "Voxel-based processing" step (<a href="#figure4">Figure 4</a>, left) can be thought of as one module for tissue segmentation and another one for spatial registration. An optional third module allows for the generation of ROIs and the calculation of ROI-based measures. The "Surface-based processing" step (<a href="#figure4">Figure 4</a>, right) can be thought of as one module for surface creation and another one for surface registration. An optional third module allows for the generation of surfaced-based ROIs and the calculation of ROI-based measures. As shown in <a href="#figure4">Figure 4</a>, the different modules utilize different priors, templates, and atlases. Those are briefly explained in the next paragraph.</p>
<p><u>Voxel-based processing:</u> While the final tissue segmentation in CAT is independent of tissue priors, the segmentation procedure is initialized using Tissue Probability Maps (TPMs). The standard TPMs (as provided in SPM) suffice for the vast majority of applications, and customized TPMs are only recommended for data obtained in young children. Please note that these TPMs should contain 6 classes: GM/WM/CSF and 3 background classes. For spatial registration, CAT uses Geodesic Shooting (<a href="https://doi.org/10.1016/j.neuroimage.2010.12.049">Ashburner and Friston, 2011</a>) or the older DARTEL approach (<a href="https://doi.org/10.1016/j.neuroimage.2007.07.007">Ashburner, 2007</a>) with predefined templates. Those templates are an appropriate choice for most studies and, again, sample-specific Shooting templates may only be advantageous for young children. For the voxel-based ROI analyses, CAT offers a selection of volume-based atlases in the predefined template space. Thus, any atlas-based ROI analysis requires a normalization of the individual scans to CAT's default Shooting template. On that note, the aforementioned creation and selection of a customized (rather than the predefined) DARTEL or Geodesic Shooting template will disable the third module for the voxel-based ROI analysis. </p>
<p><u>Surface-based processing:</u> During the surface registration, the cortical surfaces of the two hemispheres are registered to the Freesurfer <a href="https://surfer.nmr.mgh.harvard.edu/fswiki/FsAverage">"FsAverage" template</a> (as provided with CAT). For surface-based ROI analyses, CAT provides a selection of surface-based atlases. As the surface registration module uses the Freesurfer "FsAverage" template, surface-based ROI analyses are not impacted by any template modification during voxel-based processing. In contrast to voxel-based ROI analyses, surface-based ROI analyses can, therefore, be applied regardless of whether the predefined or customized versions of the DARTEL or Geodesic Shooting template have been used.</p>
<p><img id="figure4" width=80% alt="cat_processing_steps" src="images/cat_processing_steps.png"></p>
<figcaption><b>Figure 4:</b> Overview of CAT's major processing steps.</figcaption>
<!-- ############################# -->
<h4 id="process_details">CAT12 processing steps in detail</h4>
<p>The modules described in the previous section help understand the CAT's overall processing workflow, including its priors, templates, and atlases. Also, data processing in CAT can be separated into three main processes: (1) the initial voxel-based processing, (2) the main voxel-based processing, and (3) the surface-based processing (optional), as further detailed below and visualized in <a href="#figure5">Figure 5</a>.</p>
<p>The <em>"initial voxel-based processing"</em> begins with a spatial adaptive non-local means (SANLM) denoising filter (<a href="https://doi.org/10.1002/jmri.22003">Manjon et al., 2010</a>), which is followed by internal resampling to properly accommodate low-resolution images and anisotropic spatial resolutions. The data are then bias-corrected and affine-registered (to further improve the outcomes of the following steps) followed by the standard SPM "unified segmentation" (<a href="https://doi.org/10.1016/j.neuroimage.2005.02.018">Ashburner and Friston, 2005</a>). The outcomes of the latter step will provide the starting estimates for the subsequent refined voxel-based processing.</p>
<p>The <em>"refined voxel-based processing"</em> uses the output from the unified segmentation and proceeds with skull-stripping of the brain. The brain is then parcellated into the left and right hemisphere, subcortical areas, and the cerebellum. Furthermore, local white matter hyperintensities are detected (to be later accounted for during the spatial normalization and cortical thickness estimation). Subsequently, a local intensity transformation of all tissue classes is performed, which is particularly helpful to reduce the effects of higher gray matter intensities in the motor cortex, basal ganglia, or occipital lobe before the final adaptive maximum a posteriori (AMAP) segmentation. This final AMAP segmentation step (<a href="https://doi.org/10.1109/42.563663">Rajapakse et al., 1997</a>), which does not rely on a priori information of the tissue probabilities, is then refined by applying a partial volume estimation (<a href="https://doi.org/10.1016/j.neuroimage.2004.05.007">Tohka et al., 2004</a>), which effectively estimates the fractional content for each tissue type per voxel. As a last default step, the tissue segments are spatially normalized to a common reference space using Geodesic Shooting (<a href="https://doi.org/10.1016/j.neuroimage.2010.12.049">Ashburner and Friston, 2011</a>) registrations.</p>
<div class="box right"><sup>*</sup> Topology defects are holes and handles that prevent the surface from being homeomorphic with a sphere.</div>
<p>Optionally, the <em>"surface-based processing"</em> will be run following the completion of the voxel-based processing steps. Here, the cortical thickness estimation and reconstruction of the central surface occur in one step using a projection-based thickness (PBT) method (<a href="https://doi.org/10.1016/j.neuroimage.2012.09.050">Dahnke et al., 2013</a>). Importantly, this PBT allows the appropriate handling of partial volume information, sulcal blurring, and sulcal asymmetries without explicit sulcus reconstruction (<a href="https://doi.org/10.1016/j.neuroimage.2012.09.050">Dahnke et al., 2013</a>). After the initial surface reconstruction, topological defects<b><sup>*</sup></b> are repaired using spherical harmonics (<a href="https://doi.org/10.1002/hbm.21095">Yotter et al., 2011a</a>). The topological correction is followed by a surface refinement, which results in the final central surface mesh. This mesh provides the basis to extract folding patterns (i.e., based on the position of mesh nodes relative to each other), where the resulting local values (e.g., absolute mean curvature) are projected onto each node. Subsequently, the individual central surfaces are spatially registered to the Freesurfer "FsAverage" template using a spherical mapping with minimal distortions (<a href="https://doi.org/10.1016/j.neuroimage.2011.02.007">Yotter et al., 2011b</a>). In the last step, the local thickness values are transferred onto the Freesurfer "FsAverage" template. While this last step is performed by default during the surface-based processing, it can be repeated to also transfer measurements of cortical folding (e.g., gyrification) as well as other measurements or data (e.g., functional or quantitative MRI) to the Freesurfer "FsAverage" template. Note that the spatial registration of cortical measurements is combined with a spatial smoothing step in CAT to prepare the data for statistical analysis.</p>
<p><img id="figure5" width=80% alt="cat_processing_pipeline" src="images/cat_processing_pipeline.png"></p>
<figcaption><b>Figure 5:</b> Schematic flowchart of CAT's processing pipeline.</figcaption>
<!-- ################################################################################# -->
<h2 id="process_data">Preprocessing data</h2>
<!-- ############################# -->
<h5 id="module1">First Module: Segment Data</h5>
<p>Please note that additional parameters for expert users are displayed in the GUI if you set the option cat.extopts.expertgui to "1" in cat_defaults.m or call cat12 by:<br>
<code>cat12("expert")</code></p>
<p>CAT12 → Preprocessing → Segment Data<br>
Parameters:</p>
<ul>
<li>Volumes <-X → Select Files → <em>[select the new files]</em> → Done
<ul>
<li>Select one volume for each subject. As the Toolbox does not support multispectral data (i.e., different imaging methods for the same brain, such as T1-, T2-, diffusion-weighted, or CT images), it is recommended to choose a T1-weighted image.</li>
<li>Importantly, the images need to be in the same orientation as the priors; you can double-check and correct them via using "Display" in the SPM menu. The priors are located in your SPM folder "SPM12 → tpm → TPM.nii")</li>
</ul></li>
<li>Split job into separate processes → <em>[use defaults or modify]</em>
<ul>
<li>To use multi-threading the CAT12 segmentation job with multiple subjects can be split into separate processes that run in the background. If you don't want to run processes in the background then set this value to 0.</li>
<li>Keep in mind that each process needs about 1.5..2GB of RAM, which should be considered to choose the appropriate number of processes.</li>
</ul></li>
<li>Options for initial SPM12 affine registration → <em>[use defaults or modify]</em>
<ul>
<li>The defaults provide a solid starting point. The SPM12 tissue probability maps (TPMs) are used for the initial spatial registration and segmentation. Alternatively, customized TPMs can be chosen (e.g. for children data) that were created with the Template-O-Matic (TOM) Toolbox. Please note that these TPMs should contain 6 classes: GM/WM/CSF and 3 background classes.</li>
</ul></li>
<li>Extended options for CAT12 segmentation → <em>[use defaults or modify]</em>
<ul>
<li>Again, the defaults provide a solid starting point. Using the extended options you can adjust specific parameters or the strength of different corrections ("0" means no correction and "0.5" is the default value that works best for a large variety of data).</li>
<li>CAT12 provides a template for the high-dimensional Shooting registration that should work for most data. However, a customized Shooting template can be selected (e.g. for children data) that was created using the Shooting toolbox. For more information on the necessary steps, see the section "Customized Shooting Template".</li>
</ul></li>
<li>Writing options → <em>[use defaults or modify]</em>
<ul>
<li>For <b>GM, and WM image volumes</b> see <a href="#reg_info">information about native, normalized, and modulated volumes</a>.
<br><u>Note:</u> The default option "Modulated normalized" results in an analysis of relative differences in regional GM volume that have to be corrected for individual brain size in the statistical analysis using total intracranial volume (TIV).</li>
<li>A <b>Bias, noise, and globally intensity corrected T1 image</b>, in which MRI inhomogeneities and noise are removed and intensities are globally normalized, can be written in normalized space. This is useful for quality control and also to create an average image of all normalized T1 images to display/overlay the results.
<br><u>Note:</u> For a basic VBM analysis use the defaults.</li>
<li>A <b>partial volume effect (PVE) label image volume</b> can also be written in normalized or native space or as a DARTEL export file. This is useful for quality control.
<br><u>Note:</u> For a basic VBM analysis use the defaults.</li>
<li>The <b>Jacobian determinant</b> for each voxel can be written in normalized space. This information can be used to do a Deformation-Based Morphometry (DBM) analysis.
<br><u>Note:</u> For a basic VBM analysis this is not needed.</li>
<li>Finally, <b>deformation fields</b> can be written. This option is useful to re-apply normalization parameters to other co-registered images (e.g. fMRI or DTI data).
<br><u>Note:</u> For a basic VBM analysis this is not needed.</li>
</ul></li>
</ul><br>
<p>The resulting files are saved in the respective subfolders and are following the CAT12 <a href="#naming">naming convention</a>.</p>
<p><b>Note:</b> If the segmentation fails, this is often due to an unsuccessful initial spatial registration. In this case, you can try to set the origin (anterior commissure) in the Display tool. Place the cursor roughly on the anterior commissure and press "Set Origin" The now displayed correction in the coordinates can be applied to the image with the button "Reorient". This procedure must be repeated for each data set individually.</p>
<!-- ############################# -->
<h5 id="module2">Second Module: Display Slices (optionally)</h5>
<p>CAT12 → Data Quality → Single Slice Display<br>
Parameters:</p>
<ul>
<li>Sample data <-X → Select Files → <em>[select the new files]</em> → Done
<ul>
<li>Select the newly written data [e.g. the "wm*" files, which are the normalized bias-corrected volumes for VBM or the resampled and smoothed surfaces for SBM]. This tool will display one horizontal slice for each subject, thus giving a good overview of whether the segmentation and normalization procedures yielded reasonable results. For example, if the native volume had artifacts or if the native volumes had a wrong orientation, the results may look odd. Solutions: Use "Check Reg" from the SPM main menu to make sure that the native images have the same orientation as the MNI Template ("SPM → templates → T1"). Adjust if necessary using "Display" from the SPM main menu.</li>
</ul></li>
<li>Proportional scaling → <em>[use defaults or modify]</em>
<ul>
<li>Check "yes", if you display T1 volumes.</li>
</ul></li>
<li>Spatial orientation</li>
<li>Show slice in mm → <em>[use defaults or modify]</em>
<ul>
<li>This module displays horizontal slices. This default setting provides a good overview.</li>
</ul></li>
</ul>
<!-- ############################# -->
<h5 id="module3">Third Module: Estimate Total Intracranial Volume (TIV)</h5>
<p>CAT12 → Statistical Analysis → Estimate TIV<br>
Parameters:</p>
<ul>
<li>XML files <-X → Select Files → <em>[select xml-files]</em> → Done
<ul>
<li>Select the xml-files in the report-folder [e.g. the "cat_*.xml"].</li>
</ul></li>
<li>Save values → TIV only
<ul>
<li>This option will save the TIV values for each data set in the same order as the selected xml-files. Optionally you can also save the global values for each tissue class, which might be interesting for further analysis, but is not recommended if you are interested in only using TIV as a covariate.</li>
</ul></li>
<li>Output file → <em>[use defaults or modify]</em></li>
</ul>
<p><b>Please note that TIV is strongly recommended as a covariate for all VBM analyses to correct different brain sizes. This step is not necessary for deformation- or surface-based data. Please also make sure that TIV does not correlate too much with your parameters of interest (please make sure that you use "Centering" with "Overall mean", otherwise the check for orthogonality in SPM sometimes does not work correctly). In this case, you should use global scaling with TIV.</b></p>
<!-- ############################# -->
<h5 id="module4">Fourth Module: Check Sample Homogeneity</h5>
<p>CAT12 → Data Quality → Check Sample Homogeneity<br>
Parameters:<br>
<ul>
<li>Data → New: Sample data <-X → Select Files → <em>[select gray matter volumes]</em> → Done
<ul>
<li>Select the newly written data [e.g. the "mwp1*" files, which are the modulated (m) normalized (w) GM segments (p1)]. It is recommended to use the unsmoothed segmentations that provide more anatomical details. This tool visualizes the correlation between the volumes using a boxplot and correlation matrices. Thus, it will help to identify outliers. Any outlier should be carefully inspected for artifacts or pre-processing errors using "Check worst data" in the GUI. If you specify different samples the mean correlation is displayed in separate boxplots for each sample.</li>
</ul></li>
<li>Load quality measures (leave empty for autom. search) → <em>[optionally select xml-files with quality measures]</em>
<ul>
<li>Optionally select the xml-files that are saved for each data set. These files contain useful information about some estimated quality measures that can be also used for checking sample homogeneity. Please note, that the order of the xml-files must be the same as the other data files. Leave empty for automatically searching for these xml-files.</li>
</ul></li>
<li>Global scaling with TIV <em>[use defaults or modify]</em>
<ul>
<li>This option is to correct quartic mean Z-scores for TIV by global scaling. It is only meaningful for VBM data.</li>
</ul></li>
<li>Nuisance → <em>[enter nuisance variables if applicable]</em>
<ul>
<li>For each nuisance variable which you want to remove from the data before calculating the correlation, select "New: Nuisance" and enter a vector with the respective variable for each subject (e.g. age in years). All variables have to be entered in the same order as the respective volumes. You can also type "spm_load" to upload a *txt file with the covariates in the same order as the volumes. A potential nuisance parameter can be TIV if you check segmented data with the default modulation.</li>
</ul></li>
</ul>
<br><p>A window opens with a plot in which the quartic mean Z-scores for all volumes is displayed. Higher quartic mean Z-scores values mean that your data are less similar to each other. The larger the quartic mean Z-score the more deviant is this data from the sample mean. The reason we apply a power of 4 to the z-score (quartic) is to give outliers a greater weight and make them more obvious in the plot. If you click in the plot, the corresponding data is displayed in the lower right corner and allow a closer look. The slider below the image changes the displayed slice. The pop-up menus in the top right-hand corner provide more options. Here you can select other measures that are displayed in the boxplot (e.g. optional quality measures such as noise, bias, weighted overall image quality if these values were loaded). Finally, most deviating data can be displayed in the SPM graphics window to check the data more closely. For surfaces, two additional measures are available (if you have loaded the xml-files ) to utilize the quality check of the surface extraction. The Euler number gives you an idea about the number of topology defects, while the defect size indicates how many vertices of the surface are affected by topology defects. Because topology defects are mainly caused by noise and other image artifacts (e.g. motion) this gives you an impression about the potential quality of your extracted surface.</p>
<p>The boxplot in the SPM graphics window displays the quartic mean Z-score values for each subject and indicates the homogeneity of your sample. A high quartic mean Z-score in the boxplot does not always mean that this volume is an outlier or contains an artifact. If there are no artifacts in the image and if the image quality is reasonable you don't have to exclude this volume from the sample. This tool is intended to utilize the process of quality checking and there are no clear criteria defined to exclude a volume only based on the quartic mean Z-score value. However, volumes with a noticeable higher quartic mean Z-score (e.g. above two standard deviations) are indicated and should be checked more carefully. The same holds for all other measures. Data with deviating values have not necessarily to be excluded from the sample, but these data should be checked more carefully with the "Check most deviating data" option. The order of the most deviating data is changed for every measure separately. Thus, the check for the most deviating data will give you a different order for mean correlation and other measures which allows you to judge your data quality using different features.</p>
<p>If you have loaded quality measures, you can also display the ratio between weighted overall image quality (IQR) and quartic mean Z-score. These two are the most important measures for assessing image quality. Mean Z-score measures the homogeneity of your data used for statistical analysis and is therefore a measure of image quality <b>after</b> pre-processing. Data that deviate from your sample increase variance and therefore minimize effect size and statistical power. The weighted overall image quality, on the other hand, combines measurements of noise and spatial resolution of the images <b>before</b> pre-processing. Although CAT12 uses effective noise-reduction approaches (e.g. spatial adaptive non-local means filter) pre-processed images are also affected and should be checked.</p>
<p>The product between IQR and quartic mean Z-scores makes it possible to combine these two measures of image quality before and after pre-processing. A low ratio indicates good quality before and after preprocessing and means that IQR is highly rated (resulting in a low nominal number/grade) and/or quartic mean Z-score is high. Both measures contribute to the ratio and are normalized before by their standard deviation. In the respective plot, the product value is colour-coded and each point can be selected to get the filename and display the selected slice to check data more closely.</p>
<!-- ############################# -->
<h5 id="module5">Fifth Module: Smooth</h5>
<p>SPM menu → Smooth<br>
Parameters:</p>
<ul>
<li>Images to Smooth <-X → Select Files → <em>[select grey matter volumes]</em> → Done
<ul>
<li>Select the newly written data [e.g. the "mwp1" files, which are the modulated (m) normalized (w) grey matter segments (p1)].</li>
</ul></li>
<li>FWHM → <em>[use defaults or modify]</em>
<ul>
<li>6-8mm kernels are widely used for VBM. To use this setting select "edit value" and type "6 6 6" for a kernel with 6mm FWHM.</li>
</ul></li>
<li>Data Type → <em>[use defaults or modify]</em></li>
<li>Filename Prefix → <em>[use defaults or modify]</em></li>
</ul>
<!-- ###################################### -->
<h2 id="build_stats">Building the statistical model</h2>
<p>Starting with CAT12.8, a new batch mode is used to create statistical models. Now, only cross-sectional and longitudinal data are distinguished and only factorial designs and a few other necessary models are offered, but covering all possible designs. The advantage of this new batch is that it provides tailored options for VBM and many of the rather confusing and unnecessary options in "Basic Models" or "Specify 2nd-level" have now been removed and some VBM-specific options such as correction for TIV are newly provided. This should make it easier to build the statistical model in CAT12. However, the "Full factorial" design now used for cross-sectional data is a bit more complicated to use. However, you are free to use the old statistical batch from the SPM12 GUI ("Basic Models" or "Specify 2nd-level") as an alternative.</p>
<p>Here are some hints on how to relate the old models to the new factorial design. For cross-sectional VBM data, you usually have 1..n samples (which corresponds to factor levels) and optionally covariates and nuisance parameters:</p>
<table>
<tr style="font-weight: 600;">
<td colspan="1" rowspan="1" width="25%;">
<p>Number of factor levels</p>
</td>
<td colspan="1" rowspan="1" width="20%;">
<p>Number of covariates</p>
</td>
<td colspan="1" rowspan="1">
<p>Statistical Model</p>
</td>
</tr>
<tr>
<td colspan="1" rowspan="1">
<p> 1</p>
</td>
<td colspan="1" rowspan="1">
<p> 0</p>
</td>
<td colspan="1" rowspan="1">
<p>one-sample t-test</p>
</td>
</tr>
<tr>
<td colspan="1" rowspan="1">
<p> 1</p>
</td>
<td colspan="1" rowspan="1">
<p> 1</p>
</td>
<td colspan="1" rowspan="1">
<p>single regression</p>
</td>
</tr>
<tr>
<td colspan="1" rowspan="1">
<p> 1</p>
</td>
<td colspan="1" rowspan="1">
<p>>1</p>
</td>
<td colspan="1" rowspan="1">
<p>multiple regression</p>
</td>
</tr>
<tr>
<td colspan="1" rowspan="1">
<p> 2</p>
</td>
<td colspan="1" rowspan="1">
<p> 0</p>
</td>
<td colspan="1" rowspan="1">
<p>two-sample t-test</p>
</td>
</tr>
<tr>
<td colspan="1" rowspan="1">
<p>>2</p>
</td>
<td colspan="1" rowspan="1">
<p> 0</p>
</td>
<td colspan="1" rowspan="1">
<p>Anova</p>
</td>
</tr>
<tr>
<td colspan="1" rowspan="1">
<p>>1</p>
</td>
<td colspan="1" rowspan="1">
<p>>0</p>
</td>
<td colspan="1" rowspan="1">
<p>AnCova (for nuisance parameters) or Interaction (for covariates)</p>
</td>
</tr>
</table>
<!-- ############################# -->
<h4 id="stat_options">General notes and options for statistical models</h4>
<b>Please note that when using <a class="function" href="matlab:spm_jobman('interactive','','spm.tools.cat.factorial_design');">"Basic Models"</a> in the CAT12 GUI, some unnecessary options have been removed compared to the "Basic Models" in SPM12, and some useful options have been added. We therefore strongly recommend preferring to use the models in CAT12, which offer some advantages especially when defining longitudinal designs.</b><br><br>
<u>Covariates</u><br>
This option allows for the specification of covariates and nuisance variables. TIV correction should be rather defined in "Correction of TIV".<br>
Note that SPM does not make any distinction between effects of interest (including covariates) and nuisance effects. Covariates and nuisance parameters are handled in the same way in the GLM and differ only in the contrast used.
<ul>
<li>Covariates → New Covariate</li>
<li>Vector <-X enter the values of the covariates (e.g. age in years) in the same order as the respective file names or type “spm_load” to upload a *.txt file with the covariates in the same order as the volumes</li>
<li>Name <-X Specify Text (e.g. “age”)</li>
<li>Interactions → None (or optionally "With Factor 1" to account for differences between the groups)</li>
<li>Centering → Overall mean</li>
</ul>
<u>Masking</u>
<p>Absolute masking is recommended for GM or WM volume data to ensure that you don’t analyze regions that don’t contain enough GM or WM or regions outside the brain. The default value of 0.1 is a good starting point for VBM data to ensure that only the intended tissue map is analyzed. You can gradually increase this value up to "0.2" but you must make sure that your effects are not cut off.<br>
If you analyze surface data, no additional masking is necessary.
</p>
<u>Correction of TIV</u><br>
For modulated VBM data, it is strongly recommended to always correct for TIV to account for different brain sizes. There are three options for TIV correction available:<br><br>
<ol>
<li><em>ANCOVA</em>: Here, any variance that can be explained by TIV is removed from your data (i.e. in every voxel or vertex). This is the preferred option when there is no correlation between TIV and your parameter of interest. For example, a parameter of interest may be a variable in a multiple regression design that you want to relate to your structural data. If TIV is correlated with this parameter, not only will the variance explained by TIV be removed from your data, but also parts of the variance of your parameter of interest, which should be avoided.<br>
Note that for two-sample T-tests and Anova with more than 2 groups, an interaction is modeled between TIV and group that prevents any group differences from being removed from your data, even if TIV differs between your groups.<br>
Use the <a class="function" href="matlab:spm_jobman('interactive','','spm.tools.cat.tools.chec_SPM');">Check Design Orthogonality</a> option to test for any correlations between your parameters.</li>
<li><em>Global scaling</em>: If TIV correlates with your parameter of interest, you should rather use global scaling with TIV. The easiest way to define global scaling is to use <a class="function" href="matlab:spm_jobman('interactive','','spm.tools.cat.factorial_design');">"Basic Models"</a> in the CAT12 GUI. If you prefer "Basic Models" in SPM12, see <a href="#ortho">here</a> for a detailed description how to define global scaling.</li>
<li><em>No</em>: Use this option to disable TIV correction for deformation- or surface-based data where TIV correction is not necessary.</li>
</ol>
<u>Check design orthogonality and homogeneity</u><br>
In order to identify images with poor image quality or even artefacts you can use <a href="#module4">Check Sample Homogeneity</a>. The idea of this option is to check the Z-score of all files across the sample using the files that are already defined in SPM.mat.<br>
<p>The Z-score is calculated for all images and the mean for each image is plotted using a boxplot (or violin plot) and the indicated filenames. The larger the quartic mean Z-score the more deviant is this data from the sample mean. The reason we apply a power of 4 to the z-score (quartic) is to give outliers a greater weight and make them more obvious in the plot. In the plot, outliers from the sample are usually isolated from the majority of data which are clustered around the sample mean. The quartic mean Z-score is plotted at the y-axis and the x-axis reflects the data order.<br>
The advantage of re-checking sample homogeneity at this point is that the given statistical model (design) is used and potential nuisance parameters are taken into account. If you have longitudinal data, the time points of each data set are linked in the graph to indicate intra-subject data. Unsmoothed data (if available and in the same folder) are automatically detected and used. Finally, report files are used if present (i.e., if data has not been moved or the folders renamed) and quality parameters are loaded and displayed.<br>
If you have modeled TIV as a nuisance parameter, you must check whether TIV is orthogonal (in other words independent) to any other parameter of interest in your analysis (e.g. parameters you are testing for) using <a class="function" href="matlab:spm_jobman('interactive','','spm.tools.cat.tools.chec_SPM');">Check Design Orthogonality</a>. See <a href="#ortho">here</a> for a detailed description.<p>
<u>Voxel-wise covariate (experimental!)</u><br>
This experimental option allows the specification of a voxel-wise covariate. If you have more than one modality (i.e., functional and structural data), this can be used (depending on the contrast you define) to:
<ol>
<li>Remove the confounding effect of structural data (e.g. GM) on functional data or</li>
<li>Investigate the relationship (regression) between functional and structural data.</li>
</ol>
<p>In addition, an interaction can be modeled to examine whether the regression between functional and structural data differs between two groups.<br>
Please note that the saved vSPM.mat file can only be analyzed with the TFCE toolbox r221 or newer.</p>
<u>Visualization of results</u><br>
There exist different options to call results after statistical analysis:<br><br>
<ol>
<li>General function to <a class="function" href="matlab:[hReg xSPM SPM] = cat_spm_results_ui('Setup');">call results</a></li>
<li><a class="function" href="matlab:cat_vol_slice_overlay;">Slice Overlay</a> to display volume results (i.e. VBM)</li>
<li><a class="function" href="matlab:y=cat_surf_results;">Surface Overlay</a> that allows to overlay results from surfaces or volumes</li>
</ol>
Please note that options 2 an 3 require that the results are already thresholded (e.g. using <a class="function" href="matlab:spm_jobman('interactive','','spm.tools.cat.tools.T2x');">Transform and Threshold SPM-maps</a>) or were analyzed with the <a class="function" href="matlab:if exist('spm_TFCE'),spm_jobman('interactive','','spm.tools.tfce_estimate'), else cat_install_tfce, end;">TFCE</a> toolbox (preferred approach).
<!-- ############################# -->
<h5 id="two_sample">Two-sample T-Test</h5>
<p>CAT12 → Basic Models<br>
Parameters:<br>
<ul>
<li>Directory <-X → Select Files → <em>[select the working directory for your analysis]</em> → Done</li>
<li>Design → "Two-sample t-test"
<ul>
<li>Group 1 scans → Select Files → <em>[select the smoothed grey matter data for group 1; following this script these are the "smwp1" files]</em> → Done</li>
<li>Group 2 scans → Select Files → <em>[select the smoothed grey matter data for group 2]</em> → Done</li>
<li>Independence → Yes</li>
<li>Variance → Equal or Unequal</li>
</ul></li>
<li>Covariates</li>
<li>Masking
<ul>
<li>Threshold masking → Absolute → <em>[specify a value (e.g. "0.1")]</em></li>
<li>Implicit Mask → Yes</li>
<li>Explicit Mask → <None></li>
</ul></li>
<li>Correction of TIV → ANCOVA, Global scaling, or No</li>
<li>Check design orthogonality and homogeneity → <em>[use defaults]</em></li>
</ul>
<!-- ############################# -->
<h5 id="full_fact">Full factorial model (for a 2x2 Anova)</h5>
<p>CAT12 → Statistical Analysis → Basic Models</p>
Parameters:<br>
<ul>
<li>Directory <-X → Select Files → <em>[select the working directory for your analysis]</em> → Done</li>
<li>Design → "Any cross-sectional data (Full factorial)"
<ul>
<li>Factors → "New: Factor; New: Factor"<br>
Factor
<ul>
<li>Name → <em>[specify text (e.g. "sex")]</em></li>
<li>Levels → 2</li>
<li>Independence → Yes</li>
<li>Variance → Equal or Unequal</li>
</ul>
Factor
<ul>
<li>Name → <em>[specify text (e.g. "age")]</em></li>
<li>Levels → 2</li>
<li>Independence → Yes</li>
<li>Variance → Equal or Unequal</li>
</ul></ul>
<ul>
<li>Specify Cells → "New: Cell; New: Cell; New: Cell; New: Cell"<br>
Cell
<ul>
<li>Levels → <em>[specify text (e.g. "1 1")]</em></li>
<li>Scans → <em>[select files (e.g. the smoothed GM data of the young males)]</em></li>
</ul>
Cell
<ul>
<li>Levels → <em>[specify text (e.g. "1 2")]</em></li>
<li>Scans → <em>[select files (e.g. the smoothed GM data of the old males)]</em></li>
</ul>
Cell
<ul>
<li>Levels → <em>[specify text (e.g. "2 1")]</em></li>
<li>Scans → <em>[select files (e.g. the smoothed GM data of the young females)]</em></li>
</ul>
Cell
<ul>
<li>Levels → <em>[specify text (e.g. "2 2")]</em></li>
<li>Scans → <em>[select files (e.g. the smoothed GM data of the old females)]</em></li>
</ul>
</ul></li>
</ul>
<ul>
<li>Covariates* (see the text box in example for two-sample T-test)</li>
<li>Masking
<ul>
<li>Threshold Masking → Absolute → <em>[specify a value (e.g. "0.1")]</em></li>
<li>Implicit Mask → Yes</li>
<li>Explicit Mask → <None></li>
</ul>
<li>Correction of TIV → ANCOVA, Global scaling, or No</li>
<li>Check design orthogonality and homogeneity → <em>[use defaults]</em></li>
</ul>
<!-- ############################# -->
<h5 id="mult_regress">Multiple regression (linear)</h5>
<p>CAT12 → Statistical Analysis → Basic Models</p>
Parameters:<br>
<ul>
<li>Directory <-X → Select Files → <em>[select the working directory for your analysis]</em> → Done</li>
<li>Design → "Multiple Regression"
<ul>
<li>Scans → <em>[select files (e.g. the smoothed GM data of all subjects)]</em> → Done</li>
<li>Covariates → "New: Covariate"
<ul>
<li>Vector → <em>[enter the values in the same order as the respective file names of the smoothed images]</em></li>
<li>Name → <em>[specify (e.g. "age")]</em></li>
<li>Centering → Overall mean</li>
</ul></li>
<li>Intercept → Include Intercept</li>
</ul></li>
<li>Covariates</li>
<li>Masking
<ul>
<li>Threshold masking → Absolute → <em>[specify a value (e.g. "0.1")]</em></li>
<li>Implicit Mask → Yes</li>
<li>Explicit Mask → <None></li>
</ul></li>
<li>Correction of TIV → ANCOVA, Global scaling, or No</li>
<li>Check design orthogonality and homogeneity → <em>[use defaults]</em></li>
</ul>
<!-- ############################# -->
<h5 id="mult_regress_poly">Multiple regression (polynomial)</h5>
<p>To use a polynomial model, you need to estimate the polynomial function of your parameter before analyzing it. To do this, use the function cat_stat_polynomial (included with CAT12 >r1140):<br>
<code>y = cat_stat_polynomial(x,order)</code></p>
<p>where "x" is your parameter and "order" is the polynomial order (e.g. 2 for quadratic).</p>
<p><u>Example for polynomial order 2 (quadratic)</u></p>
<p>CAT12 → Statistical Analysis → Basic Models</p>
Parameters:<br>
<ul>
<li>Directory <-X → Select Files → <em>[select the working directory for your analysis]</em> → Done</li>
<li>Design → "Multiple Regression"
<ul>
<li>Scans → <em>[select files (e.g. the smoothed GM data of all subjects)]</em> → Done</li>
<li>Covariates → "New: Covariate"
<ul>
<li>Vector → <em>[specify linear term (e.g. "y(:,1)")]</em></li>
<li>Name → <em>[specify (e.g. "age-linear")]</em></li>
<li>Centering → Overall mean</li>
</ul>
<li>Covariates → "New: Covariate"
<ul>
<li>Vector → <em>[specify quadratic term (e.g. "y(:,2)")]</em></li>
<li>Name → <em>[specify (e.g. "age-quadratic")]</em></li>
<li>Centering → Overall mean</li>
</ul></li>
<li>Intercept → Include Intercept</li>
</ul></li>
<li>Covariates</li>
<li>Masking
<ul>
<li>Threshold masking → Absolute → <em>[specify a value (e.g. "0.1")]</em></li>
<li>Implicit Mask → Yes</li>
<li>Explicit Mask → <None></li>
</ul></li>
<li>Correction of TIV → ANCOVA, Global scaling, or No</li>
<li>Check design orthogonality and homogeneity → <em>[use defaults]</em></li>
</ul>
<!-- ############################# -->
<h5 id="full_fact_interact">Full factorial model (interaction)</h5>
<p>CAT12 → Statistical Analysis → Basic Models</p>
Parameters:<br>
<ul>
<li>Directory <-X → Select Files → <em>[select the working directory for your analysis]</em> → Done</li>
<li>Design → "Any cross-sectional data (Full factorial)"
<ul>
<li>Factors → "New: Factor"<br>
Factor
<ul>
<li>Name → <em>[specify text (e.g. "sex")]</em></li>
<li>Levels → 2</li>
<li>Independence → Yes</li>
<li>Variance → Equal or Unequal</li>
</ul>
</li>
<li>Specify Cells → "New: Cell; New: Cell"<br>
Cell
<ul>
<li>Levels → <em>[specify text (e.g. "1")]</em></li>
<li>Scans → <em>[select files (e.g. the smoothed GM data of the males)]</em></li>
</ul>
Cell
<ul>
<li>Levels → <em>[specify text (e.g. "2")]</em></li>
<li>Scans → <em>[select files (e.g. the smoothed GM data of the females)]</em></li>
</ul>
</ul>
</li>
<li>Covariates → "New: Covariate"
<ul>
<li>Covariate
<ul>
<li>Vector → <em>[enter the values in the same order as the respective file names of the smoothed images]</em></li>
<li>Name → <em>[specify (e.g. "age")]</em></li>
<li>Interactions → With Factor 1</li>
<li>Centering → Overall mean</li>
</ul></li>
</ul></li>
<li>Masking
<ul>
<li>Threshold Masking → Absolute → <em>[specify a value (e.g. "0.1")]</em></li>
<li>Implicit Mask → Yes</li>
<li>Explicit Mask → <None></li>
</ul>
<li>Correction of TIV → ANCOVA, Global scaling, or No</li>
<li>Check design orthogonality and homogeneity → <em>[use defaults]</em></li>
</ul>
<!-- ############################# -->
<h5 id="full_fact_poly">Full factorial model (polynomial interaction)</h5>
<p>To use a polynomial model you have to estimate the polynomial function of your parameter prior to the analysis. Use the function cat_stat_polynomial (provided with CAT12 >r1140) for that purpose:<br>
<code>y = cat_stat_polynomial(x,order)</code></p>
<p>where "x" is your parameter and "order" is the polynomial order (e.g. 2 for quadratic).</p>
<p><u>Example for polynomial order 2 (quadratic)</u></p>
<p>CAT12 → Statistical Analysis → Basic Models</p>
Parameters:<br>
<ul>
<li>Directory <-X → Select Files → <em>[select the working directory for your analysis]</em> → Done</li>
<li>Design → "Any cross-sectional data (Full factorial)"
<ul>
<li>Factors → "New: Factor"<br>
Factor
<ul>
<li>Name → <em>[specify text (e.g. "sex")]</em></li>
<li>Levels → 2</li>
<li>Independence → Yes</li>
<li>Variance → Equal or Unequal</li>
</ul>
</li>
<li>Specify Cells → "New: Cell; New: Cell"<br>
Cell
<ul>
<li>Levels → <em>[specify text (e.g. "1")]</em></li>
<li>Scans → <em>[select files (e.g. the smoothed GM data of the males)]</em></li>
</ul>
Cell
<ul>
<li>Levels → <em>[specify text (e.g. "2")]</em></li>
<li>Scans → <em>[select files (e.g. the smoothed GM data of the females)]</em></li>
</ul>
</ul>
</li>
<li>Covariates → "New: Covariate; New: Covariate"
<ul>