AutoImplant_2021/preprocessing.py at main · MWod/AutoImplant_2021 · 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
import os
import pathlib

import numpy as np
import pandas as pd

import paths as p
import utils as u

"""
Script performing the Task 1/2/3 dataset preprocessing (boundary cropping, resampling, padding to the same shape).
"""

def preprocess_task_1_training_set(input_data_folder : pathlib.Path, output_data_folder : pathlib.Path, csv_path : pathlib.Path, **preprocessing_params):
    output_spacing = preprocessing_params['output_spacing']
    output_size = preprocessing_params['output_size']
    pad_size = preprocessing_params['pad_size']
    offset = preprocessing_params['offset']

    dataframe = pd.read_csv(csv_path)
    print("Dataset size: ", len(dataframe))

    errors = list()
    for current_id, case in dataframe.iterrows():
        print("Current ID: ", current_id)
        complete_skull_path = input_data_folder / case['Complete Skull Path']
        defective_skull_path = input_data_folder / case['Defective Skull Path']
        implant_path = input_data_folder / case['Implant Path']

        complete_skull, defective_skull, implant, spacing = u.load_training_case(complete_skull_path, defective_skull_path, implant_path)
        print("Original Complete Skull Shape: ", complete_skull.shape)
        print("Original Defective Skull Shape: ", defective_skull.shape)
        print("Original Implant Shape: ", implant.shape)
        print("Initial spacing: ", spacing)

        preprocessed_complete_skull, preprocessed_defective_skull, preprocessed_implant, to_pad, internal_shape, padding = u.preprocess_training_case(defective_skull, complete_skull, implant, spacing, output_spacing, pad_size, output_size, offset)
        print("Preprocessed Complete Skull Shape: ", preprocessed_complete_skull.shape)
        print("Preprocessed Defective Skull Shape: ", preprocessed_defective_skull.shape)
        print("Preprocessed Implant Shape: ", preprocessed_implant.shape)

        recovered_complete_skull = u.postprocess_case(preprocessed_complete_skull, spacing, output_spacing, padding, to_pad, internal_shape, pad_size)
        mse = lambda a, b: np.mean((a-b)**2)
        error = mse(complete_skull, recovered_complete_skull)
        print("MSE: ", error)
        errors.append(error)

        preprocessed_complete_skull_path = output_data_folder / case['Complete Skull Path']
        preprocessed_defective_skull_path = output_data_folder / case['Defective Skull Path']
        preprocessed_implant_path = output_data_folder / case['Implant Path']

        pathlib.Path(preprocessed_complete_skull_path).parents[0].mkdir(parents=True, exist_ok=True)
        pathlib.Path(preprocessed_defective_skull_path).parents[0].mkdir(parents=True, exist_ok=True)
        pathlib.Path(preprocessed_implant_path).parents[0].mkdir(parents=True, exist_ok=True)

        u.save_volume(preprocessed_complete_skull, output_spacing, preprocessed_complete_skull_path)
        u.save_volume(preprocessed_defective_skull, output_spacing, preprocessed_defective_skull_path)
        u.save_volume(preprocessed_implant, output_spacing, preprocessed_implant_path)

    print("Mean error: ", np.mean(errors))
    print("Max error: ", np.max(errors))

def preprocess_task_1_testing_set(input_data_folder : pathlib.Path, output_data_folder : pathlib.Path, csv_path : pathlib.Path, **preprocessing_params):
    output_spacing = preprocessing_params['output_spacing']
    output_size = preprocessing_params['output_size']
    pad_size = preprocessing_params['pad_size']
    offset = preprocessing_params['offset']

    dataframe = pd.read_csv(csv_path)
    print("Dataset size: ", len(dataframe))

    errors = list()
    for current_id, case in dataframe.iterrows():
        print("Current ID: ", current_id)
        defective_skull_path = input_data_folder / case['Defective Skull Path']
        defective_skull, spacing = u.load_testing_case(defective_skull_path)
        print("Original Defective Skull Shape: ", defective_skull.shape)
        print("Initial spacing: ", spacing)

        preprocessed_defective_skull, to_pad, internal_shape, padding = u.preprocess_testing_case(defective_skull, spacing, output_spacing, pad_size, output_size, offset)
        print("Preprocessed Defective Skull Shape: ", preprocessed_defective_skull.shape)

        recovered_defective_skull = u.postprocess_case(preprocessed_defective_skull, spacing, output_spacing, padding, to_pad, internal_shape, pad_size)
        mse = lambda a, b: np.mean((a-b)**2)
        error = mse(defective_skull, recovered_defective_skull)
        print("MSE: ", error)
        errors.append(error)

        preprocessed_defective_skull_path = output_data_folder / case['Defective Skull Path']

        pathlib.Path(preprocessed_defective_skull_path).parents[0].mkdir(parents=True, exist_ok=True)

        u.save_volume(preprocessed_defective_skull, output_spacing, preprocessed_defective_skull_path)

    print("Mean error: ", np.mean(errors))
    print("Max error: ", np.max(errors))

def preprocess_task_2_testing_set(input_data_folder : pathlib.Path, output_data_folder : pathlib.Path, csv_path : pathlib.Path, **preprocessing_params):
    preprocess_task_1_testing_set(input_data_folder, output_data_folder, csv_path, **preprocessing_params)

def preprocess_task_3_training_set(input_data_folder : pathlib.Path, output_data_folder : pathlib.Path, csv_path : pathlib.Path, **preprocessing_params):
    preprocess_task_1_training_set(input_data_folder, output_data_folder, csv_path, **preprocessing_params)

def preprocess_task_3_testing_set(input_data_folder : pathlib.Path, output_data_folder : pathlib.Path, csv_path : pathlib.Path, **preprocessing_params):
    preprocess_task_1_testing_set(input_data_folder, output_data_folder, csv_path, **preprocessing_params)

def run():
    output_spacing = (1.0, 1.0, 1.0)
    output_size = (240, 200, 240)
    pad_size = 3
    offset = 35
    preprocessing_params = dict()
    preprocessing_params['output_spacing'] = output_spacing
    preprocessing_params['output_size'] = output_size
    preprocessing_params['pad_size'] = pad_size
    preprocessing_params['offset'] = offset

    preprocess_task_1_training_set(p.task_1_training_path, p.task_1_training_preprocessed_path, p.task_1_training_csv_path, **preprocessing_params)
    preprocess_task_1_training_set(p.task_1_training_path, p.task_1_training_preprocessed_path, p.task_1_validation_csv_path, **preprocessing_params)

    preprocess_task_3_training_set(p.task_3_training_path, p.task_3_training_preprocessed_path, p.task_3_training_csv_path, **preprocessing_params)
    preprocess_task_3_training_set(p.task_3_training_path, p.task_3_training_preprocessed_path, p.task_3_validation_csv_path, **preprocessing_params)


    output_spacing = (1.0, 1.0, 1.0)
    output_size = (240, 200, 240)
    pad_size = 3
    offset = 35
    preprocessing_params = dict()
    preprocessing_params['output_spacing'] = output_spacing
    preprocessing_params['output_size'] = output_size
    preprocessing_params['pad_size'] = pad_size
    preprocessing_params['offset'] = offset

    preprocess_task_1_testing_set(p.task_1_testing_path, p.task_1_testing_preprocessed_path, p.task_1_testing_csv_path, **preprocessing_params)
    preprocess_task_2_testing_set(p.task_2_testing_path, p.task_2_testing_preprocessed_path, p.task_2_testing_csv_path, **preprocessing_params)
    preprocess_task_3_testing_set(p.task_3_testing_path, p.task_3_testing_preprocessed_path, p.task_3_testing_csv_path, **preprocessing_params)


if __name__ == "__main__":
    run()