def label_cells():
parser = argparse.ArgumentParser(description='Label cells.')
parser.add_argument('input',
help="Input image filename",
metavar='<input>')
parser.add_argument('output',
help="Output image filename",
metavar='<output>')
parser.add_argument(
'-b',
'--block_size',
help="Block size for local threshold (default:{})".format(
_default_block_size),
default=_default_block_size,
type=int,
metavar='<size>')
parser.add_argument(
'--offset',
help="Offset for local threshold (default:{})".format(_default_offset),
default=_default_offset,
type=int,
metavar='<n>')
args = parser.parse_args()
image = _load_image(args.input)
wall = _extract_wall(image, args.block_size, args.offset)
cells = _label_cells(wall)
if os.path.splitext(args.output)[1] in ['.mhd', '.mha']:
mhd.write(args.output, cells)
else:
rgba = color_cells(cells)
Image.fromarray(rgba).save(args.output)
def cv(k_index, train_index, test_index):
outdir = os.path.join(base_outdir, 'k{}'.format(k_index))
os.makedirs(outdir, exist_ok=True)
for ci, cid in enumerate(tqdm.tqdm(ids_test)):
x, h = mhd.read_memmap(os.path.join(mhd_dir, cid + '.mha'))
y, _ = mhd.read(os.path.join(label_dir, cid + '.mha'))
valid_zs = [(slab_thickness // 2 <= z < len(x) - slab_thickness // 2)
and np.any(lung[z] > 0) for z in range(len(x))]
zmin = np.min(np.where(valid_zs))
zmax = np.max(np.where(valid_zs))
seeds = [SlabSeed(x, x, i) for i in range(zmin, zmax + 1)]
p = model.predict_generator(SlabGenerator(seeds,
slab_thickness,
batch_size,
False,
transform=utils.transform,
transpose=False),
max_queue_size=workers + 1,
workers=workers,
use_multiprocessing=workers > 1)
p = np.squeeze(p)
label = np.argmax(p, axis=-1).astype(np.uint8)
label = np.pad(label, ((zmin, len(x) - zmax - 1), (0, 0), (0, 0)),
mode='constant',
constant_values=0)
h['CompressedData'] = True
mhd.write(os.path.join(image_outdir, cid + '.mha'), label, h)
def extract_wall():
parser = argparse.ArgumentParser(description='Extract cell wall.')
parser.add_argument('input',
help="Input image filename",
metavar='<input>')
parser.add_argument('output',
help="Output image filename",
metavar='<output>')
parser.add_argument(
'-b',
'--block_size',
help="Block size for local threshold (default:{})".format(
_default_block_size),
default=_default_block_size,
type=int,
metavar='<size>')
parser.add_argument(
'--offset',
help="Offset for local threshold (default:{})".format(_default_offset),
default=_default_offset,
type=int,
metavar='<n>')
args = parser.parse_args()
image = _load_image(args.input)
wall = _extract_wall(image, args.block_size, args.offset)
if os.path.splitext(args.output) in ['.mhd', '.mha']:
mhd.write(args.output, wall.astype(np.uint8))
else:
cmap = np.array([[0, 0, 0, 0], [255, 0, 0, 255]]).astype(np.uint8)
Image.fromarray(cmap[wall]).save(args.output)
def modefilt():
parser = argparse.ArgumentParser(description='Mode filter.')
parser.add_argument('input', help="Input mhd filename", metavar='<input>')
parser.add_argument('-o',
'--output',
help="Output mhd filename",
metavar='<filename>',
default='filtered.mha')
parser.add_argument('--size',
help="Filter size",
metavar='<N>',
default=3,
type=int)
args = parser.parse_args()
label, h = mhd.read(args.input)
filtered = _label_filters.modefilt3(label, args.size, 0)
mhd.write(args.output, filtered, h)
def median_filter():
parser = argparse.ArgumentParser(
description='Median filter for mhd image.')
parser.add_argument('input', help="Input filename", metavar='<input>')
parser.add_argument('-o',
'--output',
help="Output filename. Defualt:%(default)s",
metavar='<output>',
default='filtered.mha')
parser.add_argument('-s',
'--size',
help="Optional argument. Default:%(default)s",
metavar='<n>',
default=3,
type=int)
args = parser.parse_args()
import mhd
from scipy.ndimage.filters import median_filter
image, h = mhd.read(args.input)
filtered = median_filter(image, args.size)
mhd.write(args.output, filtered, h)
print("epoch = {}, loss = {:.5f}".format(epochs + 1, loss.data))
optimizer.step()
out = output.data.cpu()
running_loss += loss.item()
if i % 2000 == 1999:
print('[%d,%5d] loss: %.3f' % (epochs + 1, i + 1, running_loss / 2000))
running_loss = 0.0
result_dir = os.path.join(exp_dir, 'g{0}'.format(epochs))
os.makedirs(result_dir, exist_ok=True)
# save model
net.state_dict()
torch.save(net, os.path.join(result_dir, 'unet_model_weights.h5'))
for i, test_IDs in enumerate(groups):
train_ID = [ID for ID in dataset.keys()]
for test_ID in tqdm.tqdm(train_ID, desc='Testing'):
x_test = np.fromstring(dataset[test_ID]['x'])
predict_label = (np.argmax(out, axis=0)).numpy()
mhd.write(os.path.join(result_img_dir, test_ID), predict_label)
datay = np.resize((dataset[test_ID]['y']), (2, 68, 68))
JIs[test_ID], DCs[test_ID] = evaluate(predict_label, datay)
refined = refine_labels(predict_label)
mhd.write(os.path.join(result_img_dir, 'refined_' + test_ID), refined)
refined_JIs[test_ID], refined_DCs[test_ID] = evaluate(refined, datay)
np.savetxt(os.path.join(exp_dir, 'refined_JI.csv'), np.stack([refined_JIs[ID] for ID in all_IDs]),
delimiter=",", fmt='%g')
np.savetxt(os.path.join(exp_dir, 'refined_Dice.csv'), np.stack([refined_DCs[ID] for ID in all_IDs]),
delimiter=",", fmt='%g')
history.losses,
delimiter=",",
fmt='%g')
#save model
with open(os.path.join(result_dir, 'unet_model.json'), 'w') as f:
f.write(model.to_json())
model.save_weights(os.path.join(result_dir, 'unet_model_weights.h5'))
for test_ID in test_IDs:
x_test = dataset[test_ID]['x']
predict_y = model.predict(x_test, batch_size=2, verbose=True)
predict_label = np.argmax(predict_y, axis=3).astype(np.uint8)
mhd.write(os.path.join(result_dir, test_ID + '.mhd'),
np.squeeze(predict_label.astype(np.uint8)),
header={'ElementSpacing': spacings[test_ID][::-1]})
JIs[test_ID], DCs[test_ID], ASDs[test_ID] = evaluate(
predict_label, np.squeeze(dataset[test_ID]['y']),
spacings[test_ID])
refined = postprocess.refine_label(predict_label)
mhd.write(os.path.join(result_dir, 'refined_' + test_ID + '.mhd'),
np.squeeze(refined.astype(np.uint8)),
header={'ElementSpacing': spacings[test_ID][::-1]})
refined_JIs[test_ID], refined_DCs[test_ID], refined_ASDs[
test_ID] = evaluate(refined, np.squeeze(dataset[test_ID]['y']),
spacings[test_ID])
np.savetxt(os.path.join(exp_dir, 'JI.csv'),
np.stack([JIs[ID] for ID in all_IDs]),