UTM/python/example.yaml at main · KitwareMedical/UTM · GitHub

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
#Base folde for storing results and intermeidates
workingfolder: "./example/results"
#Where should standard out be directed
progresspipe: ""
#Numbe of parallel processors to use
nparallel: 8

#atlas.Rdata file that matches atlas used for registration
#of the image data set. Required for parcel based analysis.
#Used in visualizations if avilable.
#
#The atlas R data file contains a list named atlas with
# atlas$bg.image - array of MRI image
# atlas$label.image - array of dame size as bg.image with label index for each voxel
# atlas$labels - names for each index
#(see "../Atlas/sri24/labels/atlas.Rdata used as default)
#
#if not to be used set use to OFF
atlas:
  use: ON
  file: ""

#Run setup stage
setup:
  use: ON

#Run barycenter computation
barycenters:
  use: OFF
  #Barycenter to compute, Euclidean or Wasserstein
  type: "Euclidean"
  #Where to store results relative to working folder
  folder: "Barycenter"
  #Sparsity for euclidean mean. Remove pixel with less than
  #0.3 of maximal average intensity. The number of pixels
  #in the average affects transport computation times
  sparsity: 0.3

#Run optimal transport computation
#this is required when using utm features
#but only needs to be recomputed if the input
#images changed
transport:
  use: ON
  recompute: OFF
  #folders for intermeidate compuation results
  #relative to workingfolder
  pointsfolder: "Points"
  gmrafolder: "GMRA"
  #folder for stroing final transport computatipon results
  #relative to working folder
  transportfolder: "TransportMaps"
  #Optimal transport mode [default=3]: \n
  #    0 - Balanced (equalizes source and target mass)
  #    1 - Add mass (Only allow addition of mass in soutrce or target)
  #    2 - Subtract mass (Only allow subtraction of mass in source or target)
  #    3 - Match source mass (Add or subtract in source to match target)
  #    4 - Free (Add or subtract mass anywhere)
  #    5 - Free source only (Add or subtract mass in source only)
  massbalancing: 3
  #Cost for adding mass, this adds a trade of between
  #transporting mass or adding mass
  #thus mass is only transported if it is cheaper
  #than allocation or removing mass
  #Cost effectively acts as a radius for within
  #transportation is prefered to mass allocation
  cost: 0
  degree: 2 # Should document what this is! The 2 indicates that we are using _squared_ euclidean distance for transport cost. -e

#Features to compute
features:
  folder: "Features"
  recompute: OFF
  #Convolutional optimal transport does not require transport
  #computation above
  #(https://arxiv.org/abs/1607.05816)
  #
  #(Processing/conv.transport.R)
  conv:
    use: ON
    folder: "Conv"
    #entropy regulariztion
    sigma: 1
    #how closely to match source distribution
    lambda.mean: 10
    #how closely to match target distribution
    lambda.target: 100
    #number of iterations
    iterations: 10
  #Classical vbm (smoothed input images)
  vbm:
    use: ON
    folder: "VBM"
    #smooth intensity images
    sigma: 0.5
  #unbalanced optimal transport requires transport
  #computation rom above
  #
  #(Processing
  utm:
    use: ON
    folder: "UTM"
    #Smooth mass allocation and transport feature images
    sigma: 0.5


#Analysis steps to perform
analysis:
  folder: "Analysis"
  correlation:
    use: ON
    permutations: 5000
    n.chunks: 50000
  #Run principal compnent and spatially regularized
  #component analysis see
  #(https://link.springer.com/chapter/10.1007/978-3-030-59728-3_65)
  components:
    use: OFF
    #Number of components for PCA
    ncomponents: 5
    #Number of components for spatially regularized component
    #analysis
    ncomponents.reg: 5
    #psoitive l1 regualrization term
    lambda.pos: 0.001
    #negative regularization term
    lambda.neg: 0.5
    #number of iterations
    n.iters: 5
    #step size per iteratiom
    step: 0.1
    #Not used currently
    kappa: 0.001
    #
    sigma: 1
  #Parcel based analysis requires atlas file
  parcels:
    use: ON
  #Morse-Smale parcel computation and analysis
  ms.parcels:
    use: OFF
    n.components: 40
    pLevelP: 0.2

report:
  use: OFF # Warning: Enabling this may cause legacy code in run.utm.barycenter.R to execute
  #p-value threshold for generating images in report
  threshold: 0.05
  #Extract slice for report at location x [between 0 and 1]
  axial: 0.4
  sagittal: 0.4
  coronal: 0.4