-
Notifications
You must be signed in to change notification settings - Fork 0
/
config_settings_CT.xml
138 lines (138 loc) · 10.7 KB
/
config_settings_CT.xml
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
<root>
<config>
<general>
<!-- Configuration string that may be added to distinguish different configuration -->
<config_str>tumor</config_str>
<!-- Whether calculations should be performed in 2D (True) or 3D (False). Default: False -->
<by_slice>False</by_slice>
</general>
<img_interpolate>
<!-- Controls whether interpolation is performed at all. -->
<interpolate>True</interpolate>
<!-- Sets the order of the interpolation spline. It can be 0 (nearest neighbour), 1 (linear),2 ,3 (cubic), 4, 5 -->
<spline_order>3</spline_order>
<!-- Uniform voxel sizes for interpolation. Units are defined by the headers of the image files. Every value represents the spacing that will be applied in all directions -->
<new_spacing>1.0</new_spacing>
<!-- Non-uniform voxel spacing. Requires 3 values for z, y, and x directions. -->
<new_non_iso_spacing></new_non_iso_spacing>
<!-- Anti-aliasing filter to reduce aliasing artefacts when down-sampling. Enabled by default (True). -->
<anti_aliasing></anti_aliasing>
<!-- Smoothing parameter beta for anti-aliasing filter. Must be in range (0.0, 1.0]. At 1.0 no anti-aliasing is performed. Default is 0.95. -->
<smoothing_beta>0.97</smoothing_beta>
</img_interpolate>
<roi_interpolate>
<!-- Sets the order of the interpolation spline. It can be 0 (nearest neighbour), 1 (linear), 2 (cubic), 3, 4, 5. Nearest neighbour or linear interpolation (default) are recommended. -->
<spline_order>1</spline_order>
<!-- Threshold for ROIS with partial volumes after interpolation. Default: 0.5 -->
<incl_threshold></incl_threshold>
</roi_interpolate>
<vol_adapt>
<!-- Whether the image may be cut to only maintain the interesting parts. Default: False. Setting this to True speeds up calculations and saves memory. -->
<resect>True</resect>
<!-- Number of times noise is randomly added to the image. Used in noise addition image perturbations. Default: 0 -->
<noise_repetitions>1</noise_repetitions>
<!-- Manually provided noise level in intensity units. If left unset, noise is determined from the image itself. -->
<noise_level></noise_level>
<!-- Angles (in degrees) over which the image and mask are rotated. This rotation is only in the x-y (axial) plane. Used in the rotation image perturbation. Default: 0.0.-->
<rot_angles></rot_angles>
<!-- Sub-voxel translation distance fractions of the interpolation grid. This forces the interpolation grid to shift slightly and interpolate at different points. Used in translation perturbations. Default: 0.0.-->
<translate_frac>0.0,0.25,0.75</translate_frac>
<!-- Growth/shrinkage of the ROI mask. Interpretations depends on the roi_adapt_size tag. If "distance": growth/shrinkage in world dimension units. If "fraction": growth/shrinkage in volume fraction. Default: 0.0 (no changes)-->
<roi_adapt_size>0.0,-0.15,0.15</roi_adapt_size>
<!-- Type of growth/shrinkage. Can be either fraction or distance (default). fraction is used in the volume growth/shrinkage image perturbation. -->
<roi_adapt_type>fraction</roi_adapt_type>
<!-- Limit to shrinkage of the ROI by distance-based adaptations. Fraction of the original volume. Default: 0.8 (but not used unless roi_adapt_size!=0.0 and roi_adapt_type=distance) -->
<eroded_vol_fract></eroded_vol_fract>
<!-- Supervoxel-based contour randomisation repetitions for image perturbation. Default: 0 -->
<roi_random_rep></roi_random_rep>
<!-- There are several settings for roi bulk/rim and heterogeneity-based supervoxel selection that are not documented here. -->
</vol_adapt>
<roi_resegment>
<!-- ROI re-segmentation method for intensity-based re-segmentation. Options are threshold, range, sigma and outlier. Multiple options can be provided,
and re-segmentation will take place in the given order. "threshold" and "range" are synonyms, as well as "sigma" and "outlier".
If left unset, no re-segmentation is performed.-->
<method>threshold</method>
<!-- Intensity threshold for threshold-based re-segmentation (threshold and range). If set, requires two values for lower and upper range respectively.
The upper range value can also be nan for half-open ranges. -->
<g_thresh>-150,180</g_thresh>
<!-- Number of standard deviations for outlier-based intensity re-segmentation. 3 is a common setting. -->
<sigma></sigma>
</roi_resegment>
<feature_extr>
<!-- Discretisation algorithm. Can be none (default), fixed_bin_number and/or fixed_bin_size. Combinations are possible.-->
<discr_method>fixed_bin_number</discr_method>
<!-- Bin width (in intensity units) for the fixed_bin_size algorithm. Multiple values are possible. -->
<discr_bin_width></discr_bin_width>
<!-- Bin number for the fixed_bin_number algorithm. Multiple values are possible. -->
<discr_n_bins>32</discr_n_bins>
<!-- Discretisation algorithm for the intensity-volume histogram. Can be none, fixed_bin_number and/or fixed_bin_size. If unset, the image modality determines the algorithm.-->
<ivh_discr_method>fixed_bin_number</ivh_discr_method>
<!-- Bin number for the fixed_bin_number algorithm. Default: 1000 -->
<ivh_discr_n_bins></ivh_discr_n_bins>
<!-- Bin width (in intensity units) for the fixed_bin_size algorithm. -->
<ivh_discr_bin_width></ivh_discr_bin_width>
<!-- Distance (in voxels) for GLCM for determining the neighbourhood. -->
<glcm_dist>1.0</glcm_dist>
<!-- Calculate GLCM in 2D (2d) 2.5D (2.5d) or 3D (3d). Default: same as general/by_slice. -->
<glcm_spatial_method>3D</glcm_spatial_method>
<!-- How to treat the texture matrices. Can be average (features are calculated from all matrices, then averaged), slice_merge (matrices in the same slice are combined,
features calculated and then averaged), dir_merge (matrices with the same direction are combined, features calculated and then averaged) and/or vol_merge (all matrices are combined and a single feature is calculated). -->
<glcm_merge_method>vol_merge</glcm_merge_method>
<!-- Calculate GLRLM in 2d, 2.5d or 3d. Default: same as general/by_slice. -->
<glrlm_spatial_method>3D</glrlm_spatial_method>
<!-- How to treat the texture matrices. Can be average (features are calculated from all matrices, then averaged), slice_merge (matrices in the same slice are combined,
features calculated and then averaged) and/or vol_merge (all matrices are combined and a single feature is calculated). -->
<glrlm_merge_method>vol_merge</glrlm_merge_method>
<!-- Calculate GLSZM in 2d, 2.5d or 3d. Default: same as general/by_slice. -->
<glszm_spatial_method>3D</glszm_spatial_method>
<!-- Calculate GLDZM in 2d, 2.5d or 3d. Default: same as general/by_slice. -->
<gldzm_spatial_method>3D</gldzm_spatial_method>
<!-- Calculate NGTDM in 2d, 2.5d or 3d. Default: same as general/by_slice. -->
<ngtdm_spatial_method>3D</ngtdm_spatial_method>
<!-- Distance (in voxels) for NGLDM for determining the neighbourhood. -->
<ngldm_dist>1.8</ngldm_dist>
<!-- Difference level (alpha) for NGLDM -->
<ngldm_diff_lvl>0.0</ngldm_diff_lvl>
<!-- Calculate NGLDM in 2d, 2.5d or 3d. Default: same as general/by_slice. -->
<ngldm_spatial_method>3D</ngldm_spatial_method>
</feature_extr>
<img_transform>
<!-- General flag for image transformation. Default: False -->
<perform_img_transform>TRUE</perform_img_transform>
<!-- Boundary conditions, i.e. how to calculate the filter response for voxels that are less than the filter width removed from the volume edge. -->
<boundary_condition></boundary_condition>
<!-- Spatial filters to apply. Options: wavelet, laplacian_of_gaussian, laws, mean -->
<spatial_filters>laplacian_of_gaussian</spatial_filters>
<!-- Wavelet filter to be used. The name must match a wavelet implemented in pyWavelets. -->
<wavelet_fam></wavelet_fam>
<!-- Create rotationally invariant wavelet response maps; True or False -->
<wavelet_rot_invar></wavelet_rot_invar>
<!-- Create stationary wavelets; True (default) or False (decimation will take place) -->
<wavelet_stationary></wavelet_stationary>
<!-- Laplacian of Gaussian sigma (in world space units). This specifies the width of the Gaussian filter by its standard deviation. -->
<log_sigma>5.0,3.0,2.0,1.0,0.5</log_sigma>
<!-- Number of standard deviations to allow before truncating the Gaussian filter -->
<log_sigma_truncate></log_sigma_truncate>
<!-- Whether to average over laplacian of gaussian images for different scales; True or False -->
<log_average>True</log_average>
<!-- Calculate laws texture energy image, or just the response map; True (default) or False -->
<laws_calculate_energy>True</laws_calculate_energy>
<!-- Create rotationally invariant laws kernel response maps or energy images; True (default) or False -->
<laws_rot_invar>False</laws_rot_invar>
<!-- Calculate specific laws kernels; all (default) or specific combinations, e.g. L5S5E5, E5E5E5-->
<laws_kernel></laws_kernel>
<!-- Delta for chebyshev distance between center voxel and neighbourhood boundary used to calculate texture energy: integer, default: 7-->
<laws_delta></laws_delta>
<!-- Filter size for mean filter. -->
<mean_filter_size></mean_filter_size>
</img_transform>
<deep_learning>
<!--Expected size of image and mask. If 2 values are provided, cropping is performed on the x-y plane. If 3 values are provided, cropping is performed along every direction, in z, y, x order-->
<expected_size></expected_size>
<!--Type of image intensity normalisation performed. Can be one of none, range, standardisation.-->
<normalisation></normalisation>
<!--Image intensity range. Intensities outside this range receive the nearest valid value-->
<intensity_range></intensity_range>
</deep_learning>
</config>
</root>