-
Notifications
You must be signed in to change notification settings - Fork 4
/
evaluation.py
158 lines (120 loc) · 5.76 KB
/
evaluation.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
from pathlib import Path
from pandas import merge, DataFrame
import json
from sklearn import metrics
import numpy as np
import math
from evalutils.io import CSVLoader
_VAL_NUMBER_CASES = 55
_TEST_NUMBER_CASES = 105
INPUT_DIRECTORY="/input"
OUTPUT_DIRECTORY="/output"
NUMBER_CASES = _TEST_NUMBER_CASES
def micro_specificity_calculation(y_true, y_pred, classes):
TN_total = 0
FP_total = 0
for c in classes:
# Consider the actual class as positive and the remaining classes as negative
y_true_class = (y_true == c)
y_pred_class = (y_pred == c)
# Calculate confusion matrix
tn, fp, fn, tp = metrics.confusion_matrix(y_true_class, y_pred_class).ravel()
# Accumulate true negatives and false positives
TN_total += tn
FP_total += fp
# Calculate micro-averaged specificity
specificity_micro = TN_total / (TN_total + FP_total)
return specificity_micro
def write_metrics(*, metrics):
# Write a json document that is used for ranking results on the leaderboard
with open("/output/metrics.json", "w") as f:
f.write(json.dumps(metrics))
def main():
path_ground_truth = Path(__file__).parent / 'ground-truth' / 'reference.csv'
path_submission = sorted(Path(INPUT_DIRECTORY).glob('*.csv'))[0]
file_loader = CSVLoader()
ground_truth = DataFrame(file_loader.load(fname=path_ground_truth))
submission = DataFrame(file_loader.load(fname=path_submission))
if len(submission) != NUMBER_CASES:
raise RuntimeError(f"{NUMBER_CASES} cases were expected in submission file and {len(submission)} "
f"were submitted")
cases = merge(
left=ground_truth,
right=submission,
indicator=True,
how="outer",
suffixes=("_ground_truth", "_prediction"),
on="case"
)
# Tumor pathological category (T)
y_test = cases["t_ground_truth"]
y_pred_prob_t1_t2 = cases["t1_t2_prob"]
y_pred_prob_t3 = cases["t3_prob"]
y_pred_prob_t4a = cases["t4a_prob"]
y_pred_prob_t4b = cases["t4b_prob"]
y_pred_prob = np.column_stack((y_pred_prob_t1_t2, y_pred_prob_t3, y_pred_prob_t4a, y_pred_prob_t4b))
auc_t = metrics.roc_auc_score(y_test, y_pred_prob, multi_class='ovr')
micro_sensitivity_t = metrics.recall_score(cases["t_ground_truth"],
cases["t_prediction"],
average='micro')
micro_specificity_t = micro_specificity_calculation(cases["t_ground_truth"],
cases["t_prediction"],
classes=[0, 1, 2, 3])
balanced_accuracy_t = metrics.balanced_accuracy_score(cases["t_ground_truth"],
cases["t_prediction"])
score_t = 0.4 * auc_t + 0.2 * micro_sensitivity_t + 0.2 * micro_specificity_t + 0.2 * balanced_accuracy_t
# Regional nodes pathological category (N)
y_test = cases["n_ground_truth"]
y_pred_prob_n0 = cases["n0_prob"]
y_pred_prob_n1 = cases["n1_prob"]
y_pred_prob_n2 = cases["n2_prob"]
y_pred_prob = np.column_stack(
(y_pred_prob_n0, y_pred_prob_n1, y_pred_prob_n2))
auc_n = metrics.roc_auc_score(y_test, y_pred_prob, multi_class='ovr')
micro_sensitivity_n = metrics.recall_score(cases["n_ground_truth"],
cases["n_prediction"],
average='micro')
micro_specificity_n = micro_specificity_calculation(cases["n_ground_truth"],
cases["n_prediction"],
classes=[0, 1, 2])
balanced_accuracy_n = metrics.balanced_accuracy_score(cases["n_ground_truth"],
cases["n_prediction"])
score_n = 0.4 * auc_n + 0.2 * micro_sensitivity_n + 0.2 * micro_specificity_n + 0.2 * balanced_accuracy_n
# Metastasis pathological category (M)
fpr, tpr, _ = metrics.roc_curve(cases["m_ground_truth"], cases["m1_prob"],
pos_label=1) # positive class is 1; negative class is 0
auc_m = metrics.auc(fpr, tpr)
if math.isnan(auc_m):
auc_m = 0
balanced_accuracy_m = metrics.balanced_accuracy_score(cases["m_ground_truth"],
cases["m_prediction"])
f1_score_m = metrics.f1_score(cases["m_ground_truth"], cases["m_prediction"])
sensitivity_m = metrics.recall_score(cases["m_ground_truth"], cases["m_prediction"])
tn, fp, fn, tp = metrics.confusion_matrix(cases["m_ground_truth"],
cases["m_prediction"]).ravel()
specificity_m = tn / (tn + fp)
score_m = 0.4 * auc_m + 0.2 * sensitivity_m + 0.2 * specificity_m + 0.2 * balanced_accuracy_m
# Final score
score = 0.4 * score_t + 0.3 * score_n + 0.3 * score_m
final_metrics = dict()
final_metrics["aggregates"] = {
"score_t": score_t,
"auc_t": auc_t,
"balanced_accuracy_t": balanced_accuracy_t,
"sensitivity_t": micro_sensitivity_t,
"specificity_t": micro_specificity_t,
"score_n": score_n,
"auc_n": auc_n,
"balanced_accuracy_n": balanced_accuracy_n,
"sensitivity_n": micro_sensitivity_n,
"specificity_n": micro_specificity_n,
"score_m": score_m,
"auc_m": auc_m,
"balanced_accuracy_m": balanced_accuracy_m,
"sensitivity_m": sensitivity_m,
"specificity_m": specificity_m,
"score": score
}
write_metrics(metrics=final_metrics)
if __name__ == "__main__":
main()