import sys, os
import keras.backend.tensorflow_backend as ktf
import keras
import tensorflow as tf
import pandas as pd
import numpy as np
import time
import datetime
from scipy import interp
from datetime import datetime
import datetime as dt
from Segmentation.model_keras import *
from Segmentation.load_datasets import load_filenames_2nd, load_data, keep_t2
from sklearn.metrics import auc, roc_curve
from glob2 import glob
import nibabel as nib
import seaborn as sns
import json
import warnings
warnings.filterwarnings("ignore")
RAND_SEED = 42
# Disable stdout
def blockPrint():
sys.stdout = open(os.devnull, 'w')
# Restore stdout
def enablePrint():
sys.stdout = sys.__stdout__
def train_model(networks, spaths, only_t2):
"""
Train specified model
Args:
networks (list): list of keras networks to train
spaths (list): list of output directories
only_t2 (bool): whether or not to only use T2 data
Returns:
"""
epochs = 600
batch_size = 20
block_size = [18, 142, 142]
oversamp = 1.0
oversamp_test = 1.0
lab_trun = 2
im_freq = 50
val_split = 0.2
test_split = 0.1
lr = 1e-4
adaptive_hist = False
# Load training data
image_base_path = '/media/matt/Seagate Expansion Drive/MR Data/MR_Images_Sarcoma'
# image_base_path = 'E:/MR Data/MR_Images_Sarcoma'
# Load training data
filenames = load_filenames_2nd(base_path=image_base_path)
nfiles = len(filenames)
# Remove T2 images
if only_t2:
filenames = keep_t2(filenames)
# Remove validation and test set
inds = np.array((range(nfiles)), dtype=int)
np.random.seed(RAND_SEED)
np.random.shuffle(inds)
# Validation data
val_inds = inds[:round(val_split*nfiles)]
val_file = [filenames[i] for i in val_inds]
# Test data
test_inds = inds[-round(test_split*nfiles):]
test_file = [filenames[i] for i in test_inds]
# Delete all data
filenames = [filename for i, filename in enumerate(filenames) if i not in
list(val_inds) + list(test_inds)]
# Load data
print('Loading data')
x, y, orig_size = load_data(filenames, block_size, oversamp,
lab_trun, adaptive_hist)
# val_file = val_file[:1]
x_val, Y_val, orig_size_val = load_data(val_file, block_size, oversamp,
lab_trun, adaptive_hist)
# test_file = test_file[:1]
x_test, y_test, orig_size_test = load_data(test_file, block_size,
oversamp_test,
lab_trun, adaptive_hist)
# shuffle training data
inds = np.arange(0, x.shape[0])
np.random.seed(5)
np.random.shuffle(inds)
x = x[inds]
y = y[inds]
print('Size of training set:\t\t', x.shape)
print('Size of validation set: \t', x_val.shape)
print('Size of test set: \t\t', x_test.shape)
sz_patch = x.shape
for network, spath in zip(networks, spaths):
# Set up save path
spath = spath % datetime.strftime(datetime.now(), '%Y_%m_%d_%H-%M-%S')
if not os.path.exists(spath):
os.mkdir(spath)
# Display which network is training
_, net_name = os.path.split(spath)
if only_t2:
net_name += '_t2'
print('\n\n\n')
print('Training: %s' % net_name)
print('-' * 80 + '\n')
# Save a copy of this code
save_code(spath, os.path.realpath(__file__))
# Load model
model, opt = network(pretrained_weights=None,
input_size=(sz_patch[1],
sz_patch[2],
sz_patch[3],
sz_patch[4]),
lr=lr)
# Set up callbacks
tensorboard = keras.callbacks.TensorBoard(log_dir="logs/%s_%s"
% net_name,
write_graph=False,
write_grads=False,
write_images=False,
histogram_freq=0)
ckpoint_weights = keras.callbacks.ModelCheckpoint(os.path.join(spath, 'ModelCheckpoint.h5'),
monitor='val_loss',
verbose=1,
save_best_only=True,
save_weights_only=True,
mode='auto',
period=10)
image_recon_callback = image_callback_val(x_val, Y_val, spath,
orig_size_val,
block_size=block_size,
oversamp=oversamp_test,
lab_trun=lab_trun,
im_freq=im_freq,
batch_size=batch_size)
# Train model
model.fit(x=x, y=y,
epochs=epochs,
batch_size=batch_size,
validation_data=(x_val, Y_val),
callbacks=[tensorboard,
ckpoint_weights,
image_recon_callback
]
)
model.save(os.path.join(spath, 'Trained_model.h5'))
# Calculate best threshold from training data
threshold = training_threshold(model, spath, X=x, Y=y)
# threshold = 0.5
# Evaluate test data
test_set_3D(model, x_test, y_test, spath, orig_size_test, block_size,
oversamp_test, lab_trun, batch_size=1, threshold=threshold)
def run_model_test(spaths, only_t2):
"""
Test trained network
Args:
spaths (list): list of paths which contain the trained models
only_t2 (bool): whether or not to only use T2 data
Returns:
"""
epochs = 600
batch_size = 20
block_size = [18, 142, 142]
oversamp = 1.0
oversamp_test = 1.0
lab_trun = 2
im_freq = 50
val_split = 0.2
test_split = 0.1
lr = 2e-4
adaptive_hist = False
# Load training data
image_base_path = '/media/matt/Seagate Expansion Drive/MR Data/MR_Images_Sarcoma'
# Load training data
filenames = load_filenames_2nd(base_path=image_base_path)
nfiles = len(filenames)
# Remove all but T2 images
if only_t2:
filenames = keep_t2(filenames)
# Remove validation and test set
inds = np.array((range(nfiles)), dtype=int)
np.random.seed(RAND_SEED)
np.random.shuffle(inds)
# Validation data
val_inds = inds[:round(val_split*nfiles)]
val_file = [filenames[i] for i in val_inds]
# Test data
test_inds = inds[-round(test_split*nfiles):]
test_file = [filenames[i] for i in test_inds]
# Delete all data
filenames = [filename for i, filename in enumerate(filenames) if i not in
list(val_inds) + list(test_inds)]
# Load data
x, y, orig_size = load_data(filenames, block_size, oversamp,
lab_trun, adaptive_hist)
x_test, y_test, orig_size_test = load_data(test_file, block_size,
oversamp_test,
lab_trun, adaptive_hist)
print('Size of training set:\t\t', x.shape)
print('Size of test set: \t\t', x_test.shape)
for spath in spaths:
# Display which network is training
_, net_name = os.path.split(spath)
print('\n\n\n')
print('Testing: %s' % net_name)
print('-' * 80 + '\n')
# Load trained model
model_path = os.path.join(spath, 'Trained_model.h5')
model = keras.models.load_model(model_path,
custom_objects={'dice_loss': dice_loss,
'dice_metric': dice_metric})
# Calculate best threshold from training data
threshold = training_threshold(model, spath, X=x, Y=y)
# threshold = 0.5
# Evaluate test data
test_set_3D(model, x_test, y_test, spath, orig_size_test, block_size,
oversamp_test, lab_trun, batch_size=20, threshold=threshold, vols=len(test_file), continuous=True)
def run_model_test_best_weigts(spaths, only_t2):
epochs = 600
batch_size = 20
block_size = [18, 142, 142]
oversamp = 1.0
oversamp_test = 1.0
lab_trun = 2
im_freq = 50
val_split = 0.2
test_split = 0.1
lr = 2e-4
adaptive_hist = False
# Load training data
image_base_path = '/media/matt/Seagate Expansion Drive/MR Data/MR_Images_Sarcoma'
# Load training data
filenames = load_filenames_2nd(base_path=image_base_path)
nfiles = len(filenames)
# Remove all but T2 images
if only_t2:
filenames = keep_t2(filenames)
# Remove validation and test set
inds = np.array((range(nfiles)), dtype=int)
np.random.seed(RAND_SEED)
np.random.shuffle(inds)
# Validation data
val_inds = inds[:round(val_split*nfiles)]
val_file = [filenames[i] for i in val_inds]
# Test data
test_inds = inds[-round(test_split*nfiles):]
test_file = [filenames[i] for i in test_inds]
# Delete all data
filenames = [filename for i, filename in enumerate(filenames) if i not in
list(val_inds) + list(test_inds)]
# Load data
x, y, orig_size = load_data(filenames, block_size, oversamp,
lab_trun, adaptive_hist)
x_test, y_test, orig_size_test = load_data(test_file, block_size,
oversamp_test,
lab_trun, adaptive_hist)
print('Size of training set:\t\t', x.shape)
print('Size of test set: \t\t', x_test.shape)
for spath in spaths:
# Display which network is training
_, net_name = os.path.split(spath)
print('\n\n\n')
print('Testing: %s' % net_name)
print('-' * 80 + '\n')
# Load trained model
model_path = os.path.join(spath, 'Trained_model.h5')
model = keras.models.load_model(model_path,
custom_objects={'dice_loss': dice_loss,
'dice_metric': dice_metric})
# Load best (from validation set)
weights_file = os.path.join(spath, 'ModelCheckpoint.h5')
model.load_weights(weights_file)
# Calculate best threshold from training data
threshold = training_threshold(model, spath, X=x, Y=y)
# threshold = 0.5
# Evaluate test data
test_set_3D(model, x_test, y_test, spath, orig_size_test, block_size,
oversamp_test, lab_trun, batch_size=20, threshold=threshold, vols=len(test_file), continuous=True)
def train_networks():
"""
Sets up all networks to train
Returns:
"""
# Save locations
save_base_path = '/media/matt/Seagate Expansion Drive/MR Data/ML_Results'
# List of networks to train
networks = [cnn_model_3D_3lyr_relu_dice,
cnn_model_3D_3lyr_do_relu_dice_skip,
#cnn_model_3D_FlDense_dice,
#cnn_model_3D_3lyr_do_relu_hing,
cnn_model_3D_3lyr_do_relu_xentropy,
cnn_model_3D_3lyr_do_relu_xentropy_skip
]
spaths = ['%s_' + i.__name__ for i in networks]
spaths = [os.path.join(save_base_path, path) for path in spaths]
# Train the model
train_model(networks, spaths, only_t2=False)
# T2 - only
# List of networks to train
networks = [cnn_model_3D_3lyr_relu_dice,
cnn_model_3D_3lyr_do_relu_dice_skip,
# cnn_model_3D_FlDense_dice,
# cnn_model_3D_3lyr_do_relu_hing,
cnn_model_3D_3lyr_do_relu_xentropy,
cnn_model_3D_3lyr_do_relu_xentropy_skip
]
spaths = ['%s_t2_' + i.__name__ for i in networks]
spaths = [os.path.join(save_base_path, path) for path in spaths]
# Train the model
train_model(networks, spaths, only_t2=True)
def run_networks_test():
"""
Tests all trained networks.
Returns:
"""
# Paths to trained networks
spaths = ['/media/matt/Seagate Expansion Drive/MR Data/ML_Results/2019_11_08_14-36-46_cnn_model_3D_3lyr_relu_dice',
'/media/matt/Seagate Expansion Drive/MR Data/ML_Results/2019_11_08_21-50-21_cnn_model_3D_3lyr_do_relu_dice_skip',
'/media/matt/Seagate Expansion Drive/MR Data/ML_Results/2019_11_09_06-49-45_cnn_model_3D_3lyr_do_relu_xentropy',
'/media/matt/Seagate Expansion Drive/MR Data/ML_Results/2019_11_09_14-12-47_cnn_model_3D_3lyr_do_relu_xentropy_skip',
'/media/matt/Seagate Expansion Drive/MR Data/ML_Results/2019_11_09_23-04-28_t2_cnn_model_3D_3lyr_relu_dice',
'/media/matt/Seagate Expansion Drive/MR Data/ML_Results/2019_11_10_04-50-05_t2_cnn_model_3D_3lyr_do_relu_dice_skip',
'/media/matt/Seagate Expansion Drive/MR Data/ML_Results/2019_11_10_12-28-23_t2_cnn_model_3D_3lyr_do_relu_xentropy',
'/media/matt/Seagate Expansion Drive/MR Data/ML_Results/2019_11_10_18-43-24_t2_cnn_model_3D_3lyr_do_relu_xentropy_skip'
]
# run_model_test(spaths[:4], only_t2=False)
# run_model_test(spaths[4:], only_t2=True)
run_model_test_best_weigts(spaths[:4], only_t2=False)
run_model_test_best_weigts(spaths[4:], only_t2=True)
def load_local_test_data(filenames):
X = np.empty(shape=(0, 0, 0, 0, 0))
for file in filenames:
tmpx = nib.load(file).get_data().astype(np.float).squeeze()
try:
X = np.concatenate((X, tmpx), axis=2)
except ValueError:
X = tmpx
return X
def load_train_volumes(only_t2=False, adaptive_hist=False):
# Set up image path
image_base_path = '/media/matt/Seagate Expansion Drive/MR Data/MR_Images_Sarcoma'
# Set up data constants
block_size = [18, 142, 142]
oversamp_test = 1.0
lab_trun = 2
test_split = 0.1
val_split = 0.2
# Get filenames
filenames = load_filenames_2nd(base_path=image_base_path)
nfiles = len(filenames)
# Yield the number of sets in the generator
yield round((1 - (val_split + test_split) ) * nfiles)
if only_t2:
filenames = keep_t2(filenames)
# Remove validation and test set
inds = np.array((range(nfiles)), dtype=int)
np.random.seed(RAND_SEED)
np.random.shuffle(inds)
mask = np.ones(inds.shape, dtype=bool)
# Test data
mask[:round(val_split * nfiles)] = 0
mask[-round(test_split*nfiles):] = 0
train_files = [filenames[i] for i in train_inds]
while True:
for train_file in train_files:
X_train, Y_train, _ = load_data([train_file], block_size,
oversamp_test,
lab_trun, adaptive_hist)
yield [X_train, Y_train]
def load_models(paths):
"""
Loads a list of models
Args:
paths (list): list of paths to models (not including the filename)
Returns:
"""
model = []
for path in paths:
model_name = os.path.join(path, 'Trained_model.h5')
model.append(keras.models.load_model(model_name,
custom_objects=
{'dice_loss': dice_loss}))
return model
def read_tensorboard(path, fields, label):
"""
Reads tensorboard files and returns data requests as a pandas array
Args:
path (str): path to tensorboard file
fields (list of str): fields to return
label (str): dataset identifier
Returns:
pandas dataframe
"""
# Update fields with validation data
fields = [[i] + ['val_{}'.format(i)] for i in fields]
fields = sum(fields, [])
# Set up output dictionary
df = dict()
for field in fields:
df[field] = list()
# Read the file
for e in tf.train.summary_iterator(path):
for v in e.summary.value:
for field in fields:
if field == v.tag:
df[field].append(v.simple_value)
# Convert to pandas
df = pd.DataFrame.from_dict(df)
# Append set label
df['label'] = label
df['Epoch'] = list(range(1, len(df)+1))
return df
def training_curves(spath):
"""
Load and plot training curves.
Args:
spath (str): path on which to save output
Returns:
"""
# Location with saved curves
dat_path = ['/home/matt/Documents/SegSarcoma/logs/2019_11_08_14-36-47_2019_11_08_14-36-46_cnn_model_3D_3lyr_relu_dice',
'/home/matt/Documents/SegSarcoma/logs/2019_11_08_21-50-22_2019_11_08_21-50-21_cnn_model_3D_3lyr_do_relu_dice_skip',
'/home/matt/Documents/SegSarcoma/logs/2019_11_09_06-49-45_2019_11_09_06-49-45_cnn_model_3D_3lyr_do_relu_xentropy',
'/home/matt/Documents/SegSarcoma/logs/2019_11_09_14-12-48_2019_11_09_14-12-47_cnn_model_3D_3lyr_do_relu_xentropy_skip',
'/home/matt/Documents/SegSarcoma/logs/2019_11_09_23-04-28_2019_11_09_23-04-28_t2_cnn_model_3D_3lyr_relu_dice_t2',
'/home/matt/Documents/SegSarcoma/logs/2019_11_10_04-50-05_2019_11_10_04-50-05_t2_cnn_model_3D_3lyr_do_relu_dice_skip_t2',
'/home/matt/Documents/SegSarcoma/logs/2019_11_10_12-28-24_2019_11_10_12-28-23_t2_cnn_model_3D_3lyr_do_relu_xentropy_t2',
'/home/matt/Documents/SegSarcoma/logs/2019_11_10_18-43-24_2019_11_10_18-43-24_t2_cnn_model_3D_3lyr_do_relu_xentropy_skip_t2'
]
labels = ['DICE', 'DICE - skip', 'XEntr', 'XEntr - skip', 'Dice T2', 'Dice T2 - skip', 'XEntr T2', 'XEntr T2 - skip']
fields = ['binary_accuracy', 'dice_metric', 'loss']
# Initialize dataframe
df = pd.DataFrame()
# Load data
for dpath, label in zip(dat_path, labels):
# Get tensorboard file
file = glob(os.path.join(dpath, 'events*'))[0]
# Read data
tmp = read_tensorboard(file, fields, label)
# Append data
df = pd.concat((df, tmp))
# Save path
path = '/media/matt/Seagate Expansion Drive/MR Data/ML_Results/JournalPaper/TrainingCurves'
# Plot with and without skip connections
sns.set('paper')
sns.set_style('whitegrid')
fs = 12
lw = 2.5
# Labels
labs = ['DICE', 'DICE Skip', 'XEntropy', 'XEntropy - Skip']
colors = ['b', 'r', 'g', 'k']
snames = os.path.join(path, 'Acc.svg')
plt.close()
fig, ax = plt.subplots()
for z in range(len(labels) // 2):
inds = df['label'] == labels[z]
ax.plot(df.loc[inds]['binary_accuracy'], '-', color=colors[z], lw=lw, label=labels[z])
ax.plot(df.loc[inds]['val_binary_accuracy'], ':', color=colors[z], lw=lw, label='Validation')
ax.set_xlabel('Epoch', fontsize=fs)
ax.set_ylabel('Accuracy', fontsize=fs)
ax.set_ylim([0.9, 1.0])
# ax.set_yticks(np.linspace(0.9, 1.0, 3))
ax.legend(fontsize=fs-1)
for item in list(ax.get_xticklabels() + ax.get_yticklabels()):
item.set_fontsize(fs-2)
fig.savefig(os.path.join(path, snames))
# DICE
snames = os.path.join(path, 'Dice_metric.svg')
fig, ax = plt.subplots()
for z in range(len(labels) // 2):
inds = df['label'] == labels[z]
ax.plot(df.loc[inds]['dice_metric'], '-', color=colors[z], lw=lw, label=labels[z])
ax.plot(df.loc[inds]['val_dice_metric'], ':', color=colors[z], lw=lw, label='Validation')
ax.set_xlabel('Epoch', fontsize=fs)
ax.set_ylabel('DICE', fontsize=fs)
ax.set_ylim([0.35, 1.0])
ax.legend(fontsize=fs-1, loc=4)
for item in list(ax.get_xticklabels() + ax.get_yticklabels()):
item.set_fontsize(fs-2)
fig.savefig(os.path.join(path, snames))
# Loss
snames = os.path.join(path, 'Loss.svg')
fig, ax = plt.subplots()
for z in range(len(labels) // 2):
inds = df['label'] == labels[z]
ax.plot(df.loc[inds]['loss'], '-', color=colors[z], lw=lw, label=labels[z])
ax.plot(df.loc[inds]['val_loss'], ':', color=colors[z], lw=lw, label='Validation')
ax.set_xlabel('Epoch', fontsize=fs)
ax.set_ylabel('Loss', fontsize=fs)
ax.set_ylim([0.0, 0.65])
ax.legend(fontsize=fs-1, loc=1)
for item in list(ax.get_xticklabels() + ax.get_yticklabels()):
item.set_fontsize(fs-2)
fig.savefig(os.path.join(path, snames))
## T2 ##
snames = os.path.join(path, 'Acc_t2.svg')
plt.close()
fig, ax = plt.subplots()
for z in range(len(labels) // 2, len(labels)):
inds = df['label'] == labels[z]
ax.plot(df.loc[inds]['binary_accuracy'], '-', color=colors[z-len(labels)//2], lw=lw, label=labels[z])
ax.plot(df.loc[inds]['val_binary_accuracy'], ':', color=colors[z-len(labels)//2], lw=lw, label='Validation')
ax.set_xlabel('Epoch', fontsize=fs)
ax.set_ylabel('Accuracy', fontsize=fs)
ax.set_ylim([0.9, 1.0])
# ax.set_yticks(np.linspace(0.9, 1.0, 3))
ax.legend(fontsize=fs - 1)
for item in list(ax.get_xticklabels() + ax.get_yticklabels()):
item.set_fontsize(fs - 2)
fig.savefig(os.path.join(path, snames))
# DICE
snames = os.path.join(path, 'Dice_metric_t2.svg')
fig, ax = plt.subplots()
for z in range(len(labels) // 2, len(labels)):
inds = df['label'] == labels[z]
ax.plot(df.loc[inds]['dice_metric'], '-', color=colors[z-len(labels)//2], lw=lw, label=labels[z])
ax.plot(df.loc[inds]['val_dice_metric'], ':', color=colors[z-len(labels)//2], lw=lw, label='Validation')
ax.set_xlabel('Epoch', fontsize=fs)
ax.set_ylabel('DICE', fontsize=fs)
ax.set_ylim([0.35, 1.0])
ax.legend(fontsize=fs - 1, loc=4)
for item in list(ax.get_xticklabels() + ax.get_yticklabels()):
item.set_fontsize(fs - 2)
fig.savefig(os.path.join(path, snames))
# Loss
snames = os.path.join(path, 'Loss_t2.svg')
fig, ax = plt.subplots()
for z in range(len(labels) // 2, len(labels)):
inds = df['label'] == labels[z]
ax.plot(df.loc[inds]['loss'], '-', color=colors[z-len(labels)//2], lw=lw, label=labels[z])
ax.plot(df.loc[inds]['val_loss'], ':', color=colors[z-len(labels)//2], lw=lw, label='Validation')
ax.set_xlabel('Epoch', fontsize=fs)
ax.set_ylabel('Loss', fontsize=fs)
ax.set_ylim([0.0, 0.65])
ax.legend(fontsize=fs - 1, loc=1)
for item in list(ax.get_xticklabels() + ax.get_yticklabels()):
item.set_fontsize(fs - 2)
fig.savefig(os.path.join(path, snames))
def compare_dice():
"""
Load and make plots of Dice score for each network trained.
Returns:
"""
folders = ['/media/matt/Seagate Expansion Drive/MR Data/ML_Results/2019_11_08_14-36-46_cnn_model_3D_3lyr_relu_dice',
'/media/matt/Seagate Expansion Drive/MR Data/ML_Results/2019_11_08_21-50-21_cnn_model_3D_3lyr_do_relu_dice_skip',
'/media/matt/Seagate Expansion Drive/MR Data/ML_Results/2019_11_09_06-49-45_cnn_model_3D_3lyr_do_relu_xentropy',
'/media/matt/Seagate Expansion Drive/MR Data/ML_Results/2019_11_09_14-12-47_cnn_model_3D_3lyr_do_relu_xentropy_skip',
'/media/matt/Seagate Expansion Drive/MR Data/ML_Results/2019_11_09_23-04-28_t2_cnn_model_3D_3lyr_relu_dice',
'/media/matt/Seagate Expansion Drive/MR Data/ML_Results/2019_11_10_04-50-05_t2_cnn_model_3D_3lyr_do_relu_dice_skip',
'/media/matt/Seagate Expansion Drive/MR Data/ML_Results/2019_11_10_12-28-23_t2_cnn_model_3D_3lyr_do_relu_xentropy',
'/media/matt/Seagate Expansion Drive/MR Data/ML_Results/2019_11_10_18-43-24_t2_cnn_model_3D_3lyr_do_relu_xentropy_skip']
labels = ['DICE', 'DICE - skip', 'XEntropy', 'XEntropy - skip', 'Dice T2', 'Dice T2 - skip', 'XEntropy T2', 'Xentropy T2 - skip']
file_name = 'metrics2.txt'
df = {'Net':[], 'Threshold': [], 'Precision': [], 'Recall': [], 'ROC': [], 'DICE': [], 'VOE': []}
keys = ['Threshold', 'ROC', 'Precision', 'Recall', 'DICE', 'VOE']
for folder, label in zip(folders, labels):
file = os.path.join(folder, file_name)
df['Net'].append(label)
# Read meetrics file
with open(file, 'r') as f:
dat = f.readlines()
# Append values
for ind, k in zip(range(-7, -1), keys):
tmp = [i for i in dat[ind] if i.isdigit() or i == '.']
df[k].append(float(''.join(tmp)))
# Compute VOE
from sklearn.metrics import jaccard_similarity_score
# Convert to Pandas
df = pd.DataFrame.from_dict(df)
df = df.reindex([0, 1, 4, 5, 2, 3, 6, 7])
# Write to CSV
df.to_csv('/media/matt/Seagate Expansion Drive/MR Data/ML_Results/JournalPaper/metrics-val.csv')
fig = plt.figure(figsize=(10, 6))
ax = fig.add_axes([0.1, 0.2, 0.80, 0.47])
g = sns.barplot(x='Net', y='DICE', data=df, ax=ax, palette='Set1')
plt.xticks(rotation=45)
fig.savefig('/media/matt/Seagate Expansion Drive/MR Data/ML_Results/JournalPaper/DICE_nets.svg')
plt.close(fig)
def statistical_metrics(spaths):
epochs = 600
batch_size = 20
block_size = [18, 142, 142]
oversamp = 1.0
oversamp_test = 1.0
lab_trun = 2
im_freq = 50
val_split = 0.2
test_split = 0.1
lr = 2e-4
# Load training data
image_base_path = '/media/matt/Seagate Expansion Drive/MR Data/MR_Images_Sarcoma'
# Load training data
filenames = load_filenames_2nd(base_path=image_base_path)
nfiles = len(filenames)
# Remove all but T2 images
if only_t2:
filenames = keep_t2(filenames)
# Remove validation and test set
inds = np.array((range(nfiles)), dtype=int)
np.random.seed(RAND_SEED)
np.random.shuffle(inds)
# Validation data
val_inds = inds[:round(val_split*nfiles)]
val_file = [filenames[i] for i in val_inds]
# Test data
test_inds = inds[-round(test_split*nfiles):]
test_file = [filenames[i] for i in test_inds]
# Delete all data
filenames = [filename for i, filename in enumerate(filenames) if i not in
list(val_inds) + list(test_inds)]
# Load data
x_test, y_test, orig_size_test = load_data(test_file, block_size,
oversamp_test,
lab_trun, adaptive_hist)
print('Size of test set: \t\t', x_test.shape)
for spath in spaths:
# Display which network is training
_, net_name = os.path.split(spath)
print('\n\n\n')
print('Testing: %s' % net_name)
print('-' * 80 + '\n')
# Load trained model
model_path = os.path.join(spath, 'Trained_model.h5')
model = keras.models.load_model(model_path,
custom_objects={'dice_loss': dice_loss,
'dice_metric': dice_metric})
# Load best threshold
file = os.path.join(spath, 'metrics2.txt')
# Read meetrics file
with open(file, 'r') as f:
dat = f.readlines()
# Append values
ind = -7
tmp = [i for i in dat[ind] if i.isdigit() or i == '.']
threshold = float(''.join(tmp))
# Evaluate test data
if __name__ == '__main__':
tstart = time.time()
train_networks()
run_networks_test()
training_curves(spath='/media/matt/Seagate Expansion Drive/MR Data/ML_Results/JournalPaper/TrainingCurves')
compare_dice()
print('\tTotal time (HH:MM:SS): %s\n\n' % (str(dt.timedelta(seconds=round(time.time() - tstart)))))
