-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathpipe.py
303 lines (238 loc) · 13.6 KB
/
pipe.py
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
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Created on Nov 29 2023
Last updated Jun 18 2024
@author: caryn-geady
"""
# IMPORTS
# Change working directory to be whatever directory this script is in
import os
os.chdir(os.path.dirname(__file__))
import numpy as np, pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
from matplotlib.lines import Line2D
from scipy.stats import ks_2samp
# %%
# custom functions
import Code.misc_splitting as ms
import Code.lesion_selection as ls
import Code.lesion_aggregation as la
import Code.feature_handling as fh
import Code.survival_analysis as sa
# %% DATA LOADING
# SARC021
sarc021_radiomics = pd.read_csv('Data/SARC021/SARC021_radiomics.csv')
sarc021_clinical = pd.read_csv('Data/SARC021/SARC021_clinical.csv')
# RADCURE
radcure_radiomics = pd.read_csv('Data/RADCURE/RADCURE_radiomics.csv')
radcure_clinical = pd.read_csv('Data/RADCURE/RADCURE_clinical.csv')
# radcure_clinicalIso = radcure_clinical[np.unique(radcure_radiomics.USUBJID)[0]]
# CRLM
crlm_radiomics = pd.read_csv('Data/TCIA-CRLM/CRLM_radiomics.csv')
crlm_clinical = pd.read_csv('Data/TCIA-CRLM/CRLM_clinical.csv')
# %% ANALYSIS
# stable
uniFlag = False
numfeatures = 10
# iterable
dataName = ['crlm']
numLesions = [3]
for dat in dataName:
for num in numLesions:
# adjust methods
aggMethods = ['largest','largest+','smallest','primary','lung','VWANLrg','concat','cosine','UWA','VWA']
if num == 1:
aggMethods.remove('cosine')
if dat in ['crlm','radcure']:
aggMethods.remove('lung')
if dat in ['crlm','sarc021']:
aggMethods.remove('primary')
print('----------')
print('Minimum Lesions: ',str(num))
data_dict = {
'radcure' : [radcure_radiomics,radcure_clinical],
'sarc021' : [sarc021_radiomics,sarc021_clinical],
'crlm' : [crlm_radiomics,crlm_clinical]
}
# load the data
df_imaging, df_clinical = data_dict[dat][0],data_dict[dat][1]
train,test = ms.randomSplit(df_imaging,df_clinical,0.8,True,False)
# isolate patient with at least minLesions
id_counts = df_imaging['USUBJID'].value_counts()
valid_ids = id_counts[id_counts >= num].index
df_imaging = df_imaging[df_imaging['USUBJID'].isin(valid_ids)].reset_index()
df_clinical = df_clinical[df_clinical['USUBJID'].isin(valid_ids)].reset_index()
print('Number of patients in subgroup: ',str(len(df_clinical.USUBJID)))
print('----------')
# print univariate results
if uniFlag:
sa.univariate_CPH(df_imaging,df_clinical,mod_choice='total')
pipe_dict = {
'train' : [train,True],
'test' : [test,False]
}
func_dict = {
'largest' : [ls.selectLargestLesion, lambda x: fh.featureSelection(fh.featureReduction(x,scaleFlag=True),numFeatures=numfeatures)],
'largest+' : [ls.selectLargestLesion, lambda x: fh.featureSelection(fh.featureReduction(x,numMetsFlag=True,scaleFlag=True),numFeatures=numfeatures,numMetsFlag=True)],
'smallest' : [ls.selectSmallestLesion, lambda x: fh.featureSelection(fh.featureReduction(x,scaleFlag=True),numFeatures=numfeatures)],
'primary' : [ls.selectPrimaryTumor, lambda x: fh.featureSelection(fh.featureReduction(x,scaleFlag=True),numFeatures=numfeatures)],
'lung' : [ls.selectLargestLungLesion, lambda x: fh.featureSelection(fh.featureReduction(x,scaleFlag=True),numFeatures=numfeatures)],
'UWA' : [la.calcUnweightedAverage, lambda x: fh.featureSelection(fh.featureReduction(x,scaleFlag=True),numFeatures=numfeatures)],
'VWA' : [la.calcVolumeWeightedAverage, lambda x: fh.featureSelection(fh.featureReduction(x,scaleFlag=True),numFeatures=numfeatures)],
'VWANLrg' : [la.calcVolumeWeightedAverageNLargest, lambda x: fh.featureSelection(fh.featureReduction(x,scaleFlag=True),numFeatures=numfeatures)],
'cosine' : [la.calcCosineMetrics, lambda x: x],
'concat' : [la.concatenateNLargest, lambda x: fh.featureSelection(fh.featureReduction(x,scaleFlag=True),numFeatures=numfeatures)]
}
# ----- TRAINING -----
df_imaging_train = df_imaging[df_imaging.USUBJID.isin(pipe_dict['train'][0])].reset_index()
df_clinical_train = df_clinical[df_clinical.USUBJID.isin(pipe_dict['train'][0])].reset_index()
for aggName in aggMethods:
print(dat, ' - ', aggName)
print('feature selection')
if aggName in ['concat','VWANLrg','cosine']:
trainingSet = func_dict[aggName][1](func_dict[aggName][0](df_imaging_train,df_clinical_train,numLesions=num).drop('USUBJID',axis=1))
elif aggName == 'largest+':
trainingSet = func_dict[aggName][1](func_dict[aggName][0](df_imaging_train,df_clinical_train,numMetsFlag=True).drop('USUBJID',axis=1))
else:
trainingSet = func_dict[aggName][1](func_dict[aggName][0](df_imaging_train,df_clinical_train).drop('USUBJID',axis=1))
print('Features Selected: ',trainingSet.columns)
print('----------')
# ----- TESTING -----
df_imaging_test = df_imaging[df_imaging.USUBJID.isin(pipe_dict['test'][0])].reset_index()
df_clinical_test = df_clinical[df_clinical.USUBJID.isin(pipe_dict['test'][0])].reset_index()
if aggName in ['concat','VWANLrg','cosine']:
testingSet = func_dict[aggName][0](df_imaging_test,df_clinical_test,numLesions=num,scaleFlag=True).drop('USUBJID',axis=1)[trainingSet.columns]
elif aggName == 'largest+':
testingSet = func_dict[aggName][0](df_imaging_test,df_clinical_test,scaleFlag=True,numMetsFlag=True).drop('USUBJID',axis=1)[trainingSet.columns]
else:
testingSet = func_dict[aggName][0](df_imaging_test,df_clinical_test,scaleFlag=True).drop('USUBJID',axis=1)[trainingSet.columns]
print('Training Size: ',str(len(trainingSet)))
print('Testing Size: ',str(len(testingSet)))
print('----------')
best_params_CPH, scores_CPH = sa.CPH_bootstrap(trainingSet,aggName,'OS',pipe_dict['train'][1])
test_CPH = sa.CPH_bootstrap(testingSet,aggName,'OS',pipe_dict['test'][1],param_grid=best_params_CPH)
print('----------')
# save results to file
# training
ms.add_column_to_csv('Results/Spreadsheets/'+dat+'_min'+str(num)+'_CPH_training.csv', aggName, scores_CPH)
# testing
ms.add_column_to_csv('Results/Spreadsheets/'+dat+'_min'+str(num)+'_CPH_testing.csv', aggName, [test_CPH])
# %% PLOTTING/SAVING DATA
dataName = ['radcure','sarc021','crlm']
numLesions = [1,2,3]
# univariable results for total volume of all ROIs and OS
uni_dict = {
'radcure' : [0.626,0.632,0.636],
'crlm' : [0.589,0.585,0.588],
'sarc021' : [0.609,0.607,0.569]}
for dat in dataName:
for num in numLesions:
# load data
all_data = pd.read_csv('Results/Spreadsheets/'+dat+'_min'+str(num)+'_CPH_training.csv')
test_df = pd.read_csv('Results/Spreadsheets/'+dat+'_min'+str(num)+'_CPH_testing.csv')
if dat != 'radcure':
all_data['primary'] = np.nan
test_df['primary'] = np.nan
if dat != 'sarc021':
all_data['lung'] = np.nan
test_df['lung'] = np.nan
if num == 1:
all_data['cosine'] = np.nan
test_df['cosine'] = np.nan
all_data = all_data[['largest','largest+','smallest','primary','lung','VWANLrg','concat','cosine','UWA','VWA']]
all_data.columns = ['Largest','Largest+','Smallest','Primary','Lung','VWA N-largest','Concatenation','Cosine Similarity','UWA','VWA']
test_df = test_df[['largest','largest+','smallest','primary','lung','VWANLrg','concat','cosine','UWA','VWA']]
test_df.columns = ['Largest','Largest+','Smallest','Primary','Lung','VWA N-largest','Concatenation','Cosine Similarity','UWA','VWA']
# plotting params
my_pal = ['#4daf4a','#4daf4a','#4daf4a','#4daf4a','#4daf4a','#ff7f00','#ff7f00','#ff7f00','#377eb8','#377eb8']
plt.rcParams.update({'font.size': 18})
plt.rcParams["font.family"] = "Avenir"
plt.axvline(x=uni_dict[dat][num-1],linestyle='--',color='k')
plt.axvline(x=0.5,linestyle='--',color='lightgray')
ax = sns.violinplot(data=all_data,orient='h',palette=my_pal)
sns.stripplot(data=test_df,orient='h',edgecolor='k', linewidth=1, palette=['white'] * 4,ax=ax)
# Modify the legend
legend_elements = [Line2D([0], [0], linestyle='--', color='lightgrey', label='Random'),
Line2D([0], [0], linestyle='--', color='k', label='Total Volume'),
Line2D([0], [0], marker='s', color='w', label='Lesion Selection', markeredgecolor='k',markerfacecolor='#4daf4a', markersize=10,),
Line2D([0], [0], marker='s', color='w', label='Information from Select Lesions', markeredgecolor='k',markerfacecolor='#ff7f00', markersize=10),
Line2D([0], [0], marker='s', color='w', label='Information from All Lesions', markeredgecolor='k',markerfacecolor='#377eb8', markersize=10),
Line2D([0], [0], marker='o', color='w', label='Testing Data', markeredgecolor='k',markerfacecolor='w', markersize=8)]
plt.legend(handles=legend_elements, bbox_to_anchor=(1.05, 1), loc='upper left',fontsize=14)
plt.xlabel('Concordance Index (C-Index)')
plt.xlim([0.35,1])
plt.ylabel('Method')
plt.title(dat+' - '+str(num)+('+ lesions'))
plt.savefig('Results/Figures/'+dat+'_min'+str(num)+'_CPH.png',dpi=300,bbox_inches='tight')
plt.show()
# %% SELECTIVE PLOTTING FOR SLIDES
dataName = ['radcure','sarc021','crlm']
numLesions = [1,2,3]
# univariable results for total volume of all ROIs and OS
uni_dict = {
'radcure' : [0.626,0.632,0.636],
'crlm' : [0.589,0.585,0.588],
'sarc021' : [0.609,0.607,0.569]}
for dat in dataName:
for num in numLesions:
# load data
all_data = pd.read_csv('Results/Spreadsheets/'+dat+'_min'+str(num)+'_CPH_training.csv')
test_df = pd.read_csv('Results/Spreadsheets/'+dat+'_min'+str(num)+'_CPH_testing.csv')
if dat != 'radcure':
all_data['primary'] = np.nan
test_df['primary'] = np.nan
if dat != 'sarc021':
all_data['lung'] = np.nan
test_df['lung'] = np.nan
if num == 1:
all_data['cosine'] = np.nan
test_df['cosine'] = np.nan
all_data = all_data[['largest','concat','VWA']]
all_data.columns = ['Largest','Concatenation','Volume-Weighted\n Average']
test_df = test_df[['largest','concat','VWA']]
test_df.columns = ['Largest','Concatenation','Volume-Weighted\n Average']
# plotting params
my_pal = ['#4daf4a','#ff7f00','#377eb8']
plt.rcParams.update({'font.size': 18})
plt.rcParams["font.family"] = "Avenir"
plt.figure(figsize=(7, 3))
plt.axvline(x=uni_dict[dat][num-1],linestyle='--',color='k')
plt.axvline(x=0.5,linestyle='--',color='lightgray')
ax = sns.violinplot(data=all_data,orient='h',palette=my_pal)
sns.stripplot(data=test_df,orient='h',edgecolor='k', linewidth=1, palette=['white'] * 4,ax=ax)
# Modify the legend
legend_elements = [Line2D([0], [0], linestyle='--', color='lightgrey', label='Random'),
Line2D([0], [0], linestyle='--', color='k', label='Total Volume'),
Line2D([0], [0], marker='s', color='w', label='Lesion Selection', markeredgecolor='k',markerfacecolor='#4daf4a', markersize=10,),
Line2D([0], [0], marker='s', color='w', label='Information from Select Lesions', markeredgecolor='k',markerfacecolor='#ff7f00', markersize=10),
Line2D([0], [0], marker='s', color='w', label='Information from All Lesions', markeredgecolor='k',markerfacecolor='#377eb8', markersize=10),
Line2D([0], [0], marker='o', color='w', label='Testing Data', markeredgecolor='k',markerfacecolor='w', markersize=8)]
plt.legend(handles=legend_elements, bbox_to_anchor=(1.05, 1), loc='upper left',fontsize=14)
plt.xlabel('Concordance Index (C-Index)')
plt.xlim([0.35,1])
plt.ylabel('Method')
plt.title(dat+' - '+str(num)+('+ lesions'))
plt.savefig('Results/Figures/MOD'+dat+'_min'+str(num)+'_CPH.png',dpi=300,bbox_inches='tight')
plt.show()
# %% QUANTITATIVE STUFF FOR RESULTS / DISCUSSION
# data1 = 'sarc021'
# data2 = 'sarc021'
# num1 = 2
# num2 = 3
# file1 = data1+'_min'+str(num1)+'_CPH_training.csv'
# file2 = data2+'_min'+str(num2)+'_CPH_training.csv'
# df1 = pd.read_csv('Results/Spreadsheets/'+file1)
# df2 = pd.read_csv('Results/Spreadsheets/'+file2)
# pvals = []
# med1 = []
# med2 = []
# for c in df1.columns:
# pvals.append(ks_2samp(df1[c],df2[c]).pvalue)
# med1.append(df1[c].median())
# med2.append(df2[c].median())
# pvals
# # %%
# np.min(np.array(med2)-np.array(med1))
# np.array(med2)-np.array(med1)