Models:
Samuel Callahan
SCallahan/
ct-net-models
0
Downloads: 2
[09c10e]: / main.py

Download this file

210 lines (198 with data), 12.5 kB 

  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
#main.py



import timeit



from run_experiment import DukeCTModel

from models import custom_models_ctnet, custom_models_alternative, custom_models_ablation

from load_dataset import custom_datasets



#Note that here NUM_EPOCHS is set to 2 for the purposes of quickly demonstrating

#the code on the fake data. In all of the experiments reported in the paper,

#NUM_EPOCHS was set to 100. No model actually trained all the way to 100 epochs

#due to use of early stopping.

NUM_EPOCHS = 2



if __name__=='__main__':

    ####################################

    # CTNet-83 Model on Whole Data Set #----------------------------------------

    ####################################

    tot0 = timeit.default_timer()

    DukeCTModel(descriptor = 'CTNet83',

                custom_net = custom_models_ctnet.CTNetModel,

                custom_net_args = {'n_outputs':83},

                loss = 'bce', loss_args = {},

                num_epochs=NUM_EPOCHS, patience = 15,

                batch_size = 2, device = 'all', data_parallel = True,

                use_test_set = False, task = 'train_eval',

                old_params_dir = '',

                dataset_class = custom_datasets.CTDataset_2019_10,

                dataset_args = {'label_type_ld':'disease_new',

                                    'label_meanings':'all',

                                    'num_channels':3,

                                    'pixel_bounds':[-1000,200],

                                    'data_augment':True,

                                    'crop_type':'single',

                                    'selected_note_acc_files':{'train':'','valid':''}})

    tot1 = timeit.default_timer()

    print('Total Time', round((tot1 - tot0)/60.0,2),'minutes')

    

    ###################################

    # CTNet-9 Model on Whole Data Set #-----------------------------------------

    ###################################

    tot0 = timeit.default_timer()

    DukeCTModel(descriptor = 'CTNet9',

                custom_net = custom_models_ctnet.CTNetModel,

                custom_net_args = {'n_outputs':9},

                loss = 'bce', loss_args = {},

                num_epochs=NUM_EPOCHS, patience = 15,

                batch_size = 2, device = 'all', data_parallel = True,

                use_test_set = False, task = 'train_eval',

                old_params_dir = '',

                dataset_class = custom_datasets.CTDataset_2019_10,

                dataset_args = {'label_type_ld':'disease_new',

                                    'label_meanings':['nodule','opacity','atelectasis','pleural_effusion','consolidation','mass','pericardial_effusion','cardiomegaly','pneumothorax'],

                                    'num_channels':3,

                                    'pixel_bounds':[-1000,200],

                                    'data_augment':True,

                                    'crop_type':'single',

                                    'selected_note_acc_files':{'train':'','valid':''}})

    tot1 = timeit.default_timer()

    print('Total Time', round((tot1 - tot0)/60.0,2),'minutes')

    

    ####################################################

    # CTNet-83 Model on 2000 Train and 1000 Val Subset #------------------------

    ####################################################

    tot0 = timeit.default_timer()

    DukeCTModel(descriptor = 'CTNet83_SmallData',

                custom_net = custom_models_ctnet.CTNetModel,

                custom_net_args = {'n_outputs':83},

                loss = 'bce', loss_args = {},

                num_epochs=NUM_EPOCHS, patience = 15,

                batch_size = 2, device = 'all', data_parallel = True,

                use_test_set = False, task = 'train_eval',

                old_params_dir = '',

                dataset_class = custom_datasets.CTDataset_2019_10,

                dataset_args = {'label_type_ld':'disease_new',

                                    'label_meanings':'all',

                                    'num_channels':3,

                                    'pixel_bounds':[-1000,200],

                                    'data_augment':True,

                                    'crop_type':'single',

                                    'selected_note_acc_files':{'train':'/load_dataset/fakedata/predefined_subsets/2020-01-10-imgtrain_random2000.csv',

                                        'valid':'/load_dataset/fakedata/predefined_subsets/2020-01-10-imgvalid_a_random1000.csv'}})

    tot1 = timeit.default_timer()

    print('Total Time', round((tot1 - tot0)/60.0,2),'minutes')

    

    ######################################################################

    # Alternative Arch: BodyConv Model on 2000 Train and 1000 Val Subset #------

    ######################################################################

    tot0 = timeit.default_timer()

    DukeCTModel(descriptor = 'BodyConv_SmallData',

                custom_net = custom_models_alternative.BodyConv,

                custom_net_args = {'n_outputs':83},

                loss = 'bce', loss_args = {},

                num_epochs=NUM_EPOCHS, patience = 15,

                batch_size = 2, device = 'all', data_parallel = True,

                use_test_set = False, task = 'train_eval',

                old_params_dir = '',

                dataset_class = custom_datasets.CTDataset_2019_10,

                dataset_args = {'label_type_ld':'disease_new',

                                    'label_meanings':'all',

                                    'num_channels':3,

                                    'pixel_bounds':[-1000,200],

                                    'data_augment':True,

                                    'crop_type':'single',

                                    'selected_note_acc_files':{'train':'/load_dataset/fakedata/predefined_subsets/2020-01-10-imgtrain_random2000.csv',

                                        'valid':'/load_dataset/fakedata/predefined_subsets/2020-01-10-imgvalid_a_random1000.csv'}})

    tot1 = timeit.default_timer()

    print('Total Time', round((tot1 - tot0)/60.0,2),'minutes')

    

    ####################################################################

    # Alternative Arch: 3DConv Model on 2000 Train and 1000 Val Subset #--------

    ####################################################################

    tot0 = timeit.default_timer()

    DukeCTModel(descriptor = 'ThreeDConv_SmallData',

                custom_net = custom_models_alternative.ThreeDConv,

                custom_net_args = {'n_outputs':83},

                loss = 'bce', loss_args = {},

                num_epochs=NUM_EPOCHS, patience = 15,

                batch_size = 4, device = 'all', data_parallel = True,

                use_test_set = False, task = 'train_eval',

                old_params_dir = '',

                dataset_class = custom_datasets.CTDataset_2019_10,

                dataset_args = {'label_type_ld':'disease_new',

                                    'label_meanings':'all',

                                    'num_channels':1,

                                    'pixel_bounds':[-1000,200],

                                    'data_augment':True,

                                    'crop_type':'single',

                                    'selected_note_acc_files':{'train':'/load_dataset/fakedata/predefined_subsets/2020-01-10-imgtrain_random2000.csv',

                                        'valid':'/load_dataset/fakedata/predefined_subsets/2020-01-10-imgvalid_a_random1000.csv'}})

    tot1 = timeit.default_timer()

    print('Total Time', round((tot1 - tot0)/60.0,2),'minutes')

    

    ####################################################################

    # Ablation Study: CTNet-83 (Pool) on 2000Train and 1000 Val Subset #--------

    ####################################################################

    tot0 = timeit.default_timer()

    DukeCTModel(descriptor = 'CTNet83AblatePool_SmallData',

                custom_net = custom_models_ablation.CTNetModel_Ablate_PoolInsteadOf3D,

                custom_net_args = {'n_outputs':83},

                loss = 'bce', loss_args = {},

                num_epochs=NUM_EPOCHS, patience = 15,

                batch_size = 2, device = 'all', data_parallel = True,

                use_test_set = False, task = 'train_eval',

                old_params_dir = '',

                dataset_class = custom_datasets.CTDataset_2019_10,

                dataset_args = {'label_type_ld':'disease_new',

                                    'label_meanings':'all',

                                    'num_channels':3,

                                    'pixel_bounds':[-1000,200],

                                    'data_augment':True,

                                    'crop_type':'single',

                                    'selected_note_acc_files':{'train':'/load_dataset/fakedata/predefined_subsets/2020-01-10-imgtrain_random2000.csv',

                                        'valid':'/load_dataset/fakedata/predefined_subsets/2020-01-10-imgvalid_a_random1000.csv'}})

    tot1 = timeit.default_timer()

    print('Total Time', round((tot1 - tot0)/60.0,2),'minutes')

    

    #####################################################################

    # Ablation Study: CTNet-83 (Rand) on 2000 Train and 1000 Val Subset #-------

    #####################################################################

    tot0 = timeit.default_timer()

    DukeCTModel(descriptor = 'CTNet83AblateRand_SmallData',

                custom_net = custom_models_ablation.CTNetModel_Ablate_RandomInitResNet,

                custom_net_args = {'n_outputs':83},

                loss = 'bce', loss_args = {},

                num_epochs=NUM_EPOCHS, patience = 15,

                batch_size = 2, device = 'all', data_parallel = True,

                use_test_set = False, task = 'train_eval',

                old_params_dir = '',

                dataset_class = custom_datasets.CTDataset_2019_10,

                dataset_args = {'label_type_ld':'disease_new',

                                    'label_meanings':'all',

                                    'num_channels':3,

                                    'pixel_bounds':[-1000,200],

                                    'data_augment':True,

                                    'crop_type':'single',

                                    'selected_note_acc_files':{'train':'/load_dataset/fakedata/predefined_subsets/2020-01-10-imgtrain_random2000.csv',

                                        'valid':'/load_dataset/fakedata/predefined_subsets/2020-01-10-imgvalid_a_random1000.csv'}})

    tot1 = timeit.default_timer()

    print('Total Time', round((tot1 - tot0)/60.0,2),'minutes')

    

    ###################################################

    # CTNet-1 Model on 2000 Train and 1000 Val Subset #-------------------------

    ###################################################

    for abnormality in ['nodule', 'opacity', 'atelectasis', 'pleural_effusion',

                        'consolidation', 'mass', 'pericardial_effusion',

                        'cardiomegaly', 'pneumothorax']:

        print('\n\n\n\n********** Working on',abnormality,'**********')

        tot0 = timeit.default_timer()

        DukeCTModel(descriptor = 'CTNet-'+abnormality,

                    custom_net =  custom_models_ctnet.CTNetModel,

                    custom_net_args = {'n_outputs':1},

                    loss = 'bce', loss_args = {},

                    num_epochs=NUM_EPOCHS, patience = 15,

                    batch_size = 2, device = 'all', data_parallel = True,

                    use_test_set = False, task = 'train_eval',

                    old_params_dir = '',

                    dataset_class = custom_datasets.CTDataset_2019_10,

                    dataset_args = {'label_type_ld':'disease_new',

                                        'label_meanings':[abnormality], #can be 'all' or a list of strings

                                        'num_channels':3,

                                        'pixel_bounds':[-1000,200],

                                        'data_augment':True,

                                        'crop_type':'single',

                                        'selected_note_acc_files':{'train':'/load_dataset/fakedata/predefined_subsets/2020-01-10-imgtrain_random2000.csv',

                                            'valid':'/load_dataset/fakedata/predefined_subsets/2020-01-10-imgvalid_a_random1000.csv'}}

                    )

        tot1 = timeit.default_timer()

        print('Total Time', round((tot1 - tot0)/60.0,2),'minutes')