def get_Random_train(self, num, path='../nodule_cubes/train_data/csv/'):
data = []
for i in tqdm(range(num)):
while True:
t = random.randint(0, 299)
CT = self.traindata[t]
z = random.randint(30, CT.shape[0]-30)
y = random.randint(60, CT.shape[1]-60)
x = random.randint(60, CT.shape[2]-60)
tmp = CT[z - 16:z + 16, y - 16:y + 16, x - 16:x + 16]
if (tmp > -600).sum() > 5:
break
v_center = np.array([z, y, x], dtype='float')
v_center = v_center * self.new_spacings[self.labels[t]] + self.origions[self.labels[t]]
data.append({'seriesuid': self.labels[t],
'coordX': v_center[2],
'coordY': v_center[1],
'coordZ': v_center[0],
'probability': 0.5})
csv = save_results(path + 'annotations.csv')
csv.write(data)
def noduleFilter(self):
New_spacing = np.load(self.pkl_path + 'new_spacing.pkl')
Origion = np.load(self.pkl_path + 'origion.pkl')
csvdir = save_results(self.csv_save_path + 'annotations.csv')
result = []
for patient in enumerate(tqdm(self.mask_path)):
patient = patient[1]
nodules_masks = self.csv_data2[self.csv_data2.file == patient]
patient_name = nodules_masks.seriesuid.values[0]
self.mask = np.load(patient)
origion = Origion[patient_name]
new_spacing = New_spacing[patient_name]
for index, nodule in nodules_masks.iterrows():
nodule_center = np.array([nodule.coordZ, nodule.coordY, nodule.coordX])
v_center = np.rint((nodule_center - origion) / new_spacing)
v_center = np.array(v_center, dtype=int)
if self.mask[v_center[0], v_center[1], v_center[2]] == 0:
tmp = {}
tmp["seriesuid"] = patient_name
tmp["coordX"] = nodule["coordX"]
tmp["coordY"] = nodule["coordY"]
tmp["coordZ"] = nodule["coordZ"]
tmp["probability"] = nodule["probability"]
result.append(tmp)
csvdir.write(result)
print ('Filter csv: %5d' % (len(result)))
def save_position(self, path=None):
if path is None:
path = './' + self.data_type + '_csv/annotations.csv'
res = save_results(path)
res.write(self.position)
print 'save successfully!'
else:
negetive_num += 1
print '[%d / %d]' % (position_num, negetive_num)
return position
if __name__ == '__main__':
data = _3D_data(valpath='../nodule_cubes/val_data/')
labels, valdata, origins, new_spacings, old_spacings = data.get_val()
model = _3D_CNN_1(weights_path=weights_path)
index = range(len(valdata))
# random.shuffle(index)
# index = index[0:20]
position = prec(index, [32, 32, 32],
labels,
origins,
new_spacings,
stride=[32, 32, 32])
position = prec(index, [14, 14, 14],
labels,
origins,
new_spacings,
stride=[14, 14, 14],
position=position)
reaults = save_results(csvfile_path='./val_csv/annotations.csv')
reaults.write(position)