def __getitem__(self, idx):
imagepath = osp.join(self.filepath,'images',self.list_files[idx]+'.png')
pilimg = Image.open(imagepath).convert("L")
H, W = pilimg.width , pilimg.height
#PREPROCESSING STEP FOR ALL TRAIN, VALIDATION AND TEST INPUTS
#Fixed gamma value for
table = 255.0*(np.linspace(0, 1, 256)**0.8)
pilimg = cv2.LUT(np.array(pilimg), table)
if self.split != 'test':
labelpath = osp.join(self.filepath,'labels',self.list_files[idx]+'.npy')
label = np.load(labelpath)
label = np.resize(label,(W,H))
label = Image.fromarray(label)
if self.transform is not None:
if self.split == 'train':
if random.random() < 0.2:
pilimg = Starburst_augment()(np.array(pilimg))
if random.random() < 0.2:
pilimg = Line_augment()(np.array(pilimg))
if random.random() < 0.2:
pilimg = Gaussian_blur()(np.array(pilimg))
if random.random() < 0.4:
pilimg, label = Translation()(np.array(pilimg),np.array(label))
img = self.clahe.apply(np.array(np.uint8(pilimg)))
img = Image.fromarray(img)
if self.transform is not None:
if self.split == 'train':
img, label = RandomHorizontalFlip()(img,label)
img = self.transform(img)
if self.split != 'test':
## This is for boundary aware cross entropy calculation
spatialWeights = cv2.Canny(np.array(label),0,3)/255
spatialWeights=cv2.dilate(spatialWeights,(3,3),iterations = 1)*20
##This is the implementation for the surface loss
# Distance map for each class
distMap = []
for i in range(0, 4):
distMap.append(one_hot2dist(np.array(label)==i))
distMap = np.stack(distMap, 0)
# spatialWeights=np.float32(distMap)
if self.split == 'test':
##since label, spatialWeights and distMap is not needed for test images
return img,0,self.list_files[idx],0,0
label = MaskToTensor()(label)
return img, label, self.list_files[idx],spatialWeights,np.float32(distMap)
def test_empty(self):
a = np.zeros((1, 256, 256))
o = utils.class2one_hot(torch.Tensor(a).type(torch.float32),
C=2).numpy()
res = utils.one_hot2dist(o[0])
self.assertEqual(res.shape, (2, 256, 256))
self.assertEqual(res[1].sum(), 0)
self.assertEqual((res[0] <= 0).sum(), a.size)
def test_full_coverage(self):
a = np.zeros((1, 256, 256))
a[:, 50:60, :] = 1
o = utils.class2one_hot(torch.Tensor(a).type(torch.float32),
C=2).numpy()
res = utils.one_hot2dist(o[0])
self.assertEqual(res.shape, (2, 256, 256))
self.assertEqual((res[1] <= 0).sum(), a.sum())
self.assertEqual((res[1] > 0).sum(), (1 - a).sum())
def test_closure(self):
a = np.zeros((1, 256, 256))
a[:, 50:60, :] = 1
o = utils.class2one_hot(torch.Tensor(a).type(torch.float32),
C=2).numpy()
res = utils.one_hot2dist(o[0])
self.assertEqual(res.shape, (2, 256, 256))
neg = (res <= 0) * res
self.assertEqual(neg.sum(), (o * res).sum())
def test_border(self):
"""
Make sure the border inside the object is 0 in the distance map
"""
for l in range(3, 5):
a = np.zeros((1, 25, 25))
a[:, 3:3 + l, 3:3 + l] = 1
o = utils.class2one_hot(torch.Tensor(a).type(torch.float32),
C=2).numpy()
res = utils.one_hot2dist(o[0])
self.assertEqual(res.shape, (2, 25, 25))
border = (res[1] == 0)
self.assertEqual(border.sum(), 4 * (l - 1))
def test_max_dist(self):
"""
The max dist for a box should be at the midle of the object, +-1
"""
a = np.zeros((1, 256, 256))
a[:, 1:254, 1:254] = 1
o = utils.class2one_hot(torch.Tensor(a).type(torch.float32),
C=2).numpy()
res = utils.one_hot2dist(o[0])
self.assertEqual(res.shape, (2, 256, 256))
self.assertEqual(res[0].max(), 127)
self.assertEqual(np.unravel_index(res[0].argmax(), (256, 256)),
(127, 127))
self.assertEqual(res[1].min(), -126)
self.assertEqual(np.unravel_index(res[1].argmin(), (256, 256)),
(127, 127))
def save_slices(
ct_paths,
cbf_paths,
cbv_paths,
mtt_paths,
tmax_paths,
gt_paths,
dest_dir: Path,
shape: Tuple[int],
n_augment: int,
ct_dir: str = "ct",
cbf_dir="cbf",
cbv_dir="cbv",
mtt_dir="mtt",
tmax_dir="tmax",
gt_dir: str = "gt",
in_npy_dir="in_npy",
gt_npy_dir='gt_npy'
) -> Tuple[int, int, Dict, Tuple[float, float, float]]:
p_id: str = get_p_id(ct_paths)
assert len(
set(
map_(get_p_id, [
ct_paths, cbf_paths, cbv_paths, mtt_paths, tmax_paths, gt_paths
]))) == 1
space_dict: Dict[str, Tuple[float, float]] = {}
# Load the data
dx, dy, dz = nib.load(str(ct_paths)).header.get_zooms()
assert dx == dy
ct = np.asarray(nib.load(str(ct_paths)).dataobj)
cbf = np.asarray(nib.load(str(cbf_paths)).dataobj)
cbv = np.asarray(nib.load(str(cbv_paths)).dataobj)
mtt = np.asarray(nib.load(str(mtt_paths)).dataobj)
tmax = np.asarray(nib.load(str(tmax_paths)).dataobj)
gt = np.asarray(nib.load(str(gt_paths)).dataobj)
assert len(set(map_(np.shape, [ct, cbf, cbv, mtt, tmax, gt]))) == 1
assert ct.dtype in [np.int32], ct.dtype
assert cbf.dtype in [np.uint16], cbf.dtype
assert cbv.dtype in [np.uint16], cbv.dtype
assert mtt.dtype in [np.float64], mtt.dtype
assert tmax.dtype in [np.float64], tmax.dtype
assert gt.dtype in [np.uint8], gt.dtype
pos: int = (gt == 1).sum()
neg: int = (gt == 0).sum()
x, y, z = ct.shape
# Normalize and check data content
norm_ct = norm_arr(
ct) # We need to normalize the whole 3d img, not 2d slices
norm_cbf = norm_arr(cbf)
norm_cbv = norm_arr(cbv)
norm_mtt = norm_arr(mtt)
norm_tmax = norm_arr(tmax)
assert 0 == norm_ct.min() and norm_ct.max() == 1, (norm_ct.min(),
norm_ct.max())
assert 0 == norm_cbf.min() and norm_cbf.max() == 1, (norm_cbf.min(),
norm_cbf.max())
assert 0 == norm_cbv.min() and norm_cbv.max() == 1, (norm_cbv.min(),
norm_cbv.max())
assert 0 == norm_mtt.min() and norm_mtt.max() == 1, (norm_mtt.min(),
norm_mtt.max())
assert 0 == norm_tmax.min() and norm_tmax.max() == 1, (norm_tmax.min(),
norm_tmax.max())
one_hot_gt: Tensor = class2one_hot(torch.tensor(gt[None, ...],
dtype=torch.int64),
K=2)[0]
assert one_hot_gt.shape == (2, 256, 256, z), one_hot_gt.shape
distmap: np.ndarray = one_hot2dist(one_hot_gt.numpy(),
resolution=(dx, dy, dz),
dtype=np.float32)
save_dir_ct: Path = Path(dest_dir, ct_dir)
save_dir_cbf: Path = Path(dest_dir, cbf_dir)
save_dir_cbv: Path = Path(dest_dir, cbv_dir)
save_dir_mtt: Path = Path(dest_dir, mtt_dir)
save_dir_tmax: Path = Path(dest_dir, tmax_dir)
save_dir_gt: Path = Path(dest_dir, gt_dir)
save_dir_in_npy: Path = Path(dest_dir, in_npy_dir)
save_dir_gt_npy: Path = Path(dest_dir, gt_npy_dir)
save_dir_distmap_npy: Path = Path(dest_dir, "3d_distmap")
save_dirs = [
save_dir_ct, save_dir_cbf, save_dir_cbv, save_dir_mtt, save_dir_tmax,
save_dir_gt
]
for j in range(ct.shape[-1]):
ct_s = norm_ct[:, :, j]
cbf_s = norm_cbf[:, :, j]
cbv_s = norm_cbv[:, :, j]
mtt_s = norm_mtt[:, :, j]
tmax_s = norm_tmax[:, :, j]
gt_s = gt[:, :, j]
dist_s = distmap[:, :, :, j]
slices = [ct_s, cbf_s, cbv_s, mtt_s, tmax_s, gt_s]
assert ct_s.shape == cbf_s.shape == cbv_s.shape, mtt_s.shape == tmax_s.shape == gt_s.shape
assert gt_s.shape == dist_s[0, ...].shape, ((x, y, z), gt_s.shape,
dist_s.shape)
assert set(np.unique(gt_s)).issubset([0, 1])
# if gt_s.sum() > 0:
# print(f"{dist_s[1].min()=} {dist_s[1].max()=}")
# _, axes = plt.subplots(nrows=1, ncols=3)
# axes[0].imshow(gt_s)
# axes[0].set_title("GT")
# tmp = axes[1].imshow(dist_s[1, ...], cmap='rainbow')
# axes[1].set_title("Signed distance map")
# plt.colorbar(tmp, ax=axes[1])
# tmp = axes[2].imshow(np.abs(dist_s[1, ...]), cmap='rainbow')
# axes[2].set_title("Abs distance map")
# plt.colorbar(tmp, ax=axes[2])
# plt.show()
for k in range(n_augment + 1):
if k == 0:
to_save = slices
else:
to_save = map_(np.asarray, augment_arr(*slices))
assert to_save[0].shape == slices[0].shape, (to_save[0].shape,
slices[0].shape)
filename = f"{p_id}_{k:02d}_{j:04d}"
space_dict[filename] = (dx, dy)
for save_dir, data in zip(save_dirs, to_save):
save_dir.mkdir(parents=True, exist_ok=True)
if "gt" not in str(save_dir):
img = (data * 255).astype(np.uint8)
else:
img = data.astype(np.uint8)
with warnings.catch_warnings():
warnings.filterwarnings("ignore", category=UserWarning)
imsave(str(Path(save_dir, filename).with_suffix(".png")),
img)
multimodal = np.stack(
to_save[:-1]) # Do not include the ground truth
assert 0 <= multimodal.min() and multimodal.max() <= 1
save_dir_in_npy.mkdir(parents=True, exist_ok=True)
save_dir_gt_npy.mkdir(parents=True, exist_ok=True)
np.save(
Path(save_dir_in_npy, filename).with_suffix(".npy"),
multimodal)
np.save(
Path(save_dir_gt_npy, filename).with_suffix(".npy"),
to_save[-1])
save_dir_distmap_npy.mkdir(parents=True, exist_ok=True)
np.save(
Path(save_dir_distmap_npy, filename).with_suffix(".npy"),
dist_s)
return neg, pos, space_dict, (dx, dy, dz)
def save_slices(
flair_path,
t1_path,
gt_path,
dest_dir: Path,
shape: Tuple[int],
n_augment: int,
discard_negatives: bool,
flair_dir: str = "flair",
t1_dir="t1",
gt_dir: str = "gt",
in_npy_dir="in_npy",
gt_npy_dir='gt_npy'
) -> Tuple[int, int, Dict, Tuple[float, float, float]]:
p_id: str = get_p_id(flair_path)
assert len(set(map_(get_p_id, [flair_path, t1_path, gt_path]))) == 1
print(p_id)
space_dict: Dict[str, Tuple[float, float]] = {}
# Load the data
dx, dy, dz = nib.load(str(flair_path)).header.get_zooms()
# assert dx == dy, (dx, dy)
flair = np.asarray(nib.load(str(flair_path)).dataobj)
w, h, _ = flair.shape
x, y, z = flair.shape
t1 = np.asarray(nib.load(str(t1_path)).dataobj)
gt = np.asarray(nib.load(str(gt_path)).dataobj)
assert set(np.unique(gt)) <= set([0., 1., 2.])
pos: int = (gt == 1).sum()
neg: int = ((gt == 0) | (gt == 2)).sum()
assert len(set(map_(np.shape, [flair, t1, gt]))) == 1
assert flair.dtype in [np.float32], flair.dtype
assert t1.dtype in [np.uint16], t1.dtype
assert gt.dtype in [np.float32], gt.dtype
# Normalize and check data content
norm_flair = norm_arr(
flair) # We need to normalize the whole 3d img, not 2d slices
norm_t1 = norm_arr(t1)
norm_gt = gt.astype(np.uint8)
assert 0 == norm_flair.min() and norm_flair.max() == 1, (norm_flair.min(),
norm_flair.max())
assert 0 == norm_t1.min() and norm_t1.max() == 1, (norm_t1.min(),
norm_t1.max())
assert np.array_equal(np.unique(gt), np.unique(norm_gt))
resized_flair = resize(norm_flair, (256, 256, z),
mode='constant',
preserve_range=True,
anti_aliasing=False).astype(np.float32)
resized_t1 = resize(norm_t1, (256, 256, z),
mode='constant',
preserve_range=True,
anti_aliasing=False).astype(np.float32)
resized_gt = resize(norm_gt, (256, 256, z),
mode='constant',
preserve_range=True,
anti_aliasing=False,
order=0).astype(np.uint8)
resized_gt[np.where(
resized_gt == 2)] = 0 # Count those labels as background
# Pre-compute the 3d distance map
rx = dx * w / 256
ry = dy * h / 256
rz = dz
# print(f"{flair.shape=}")
# print(f"{(dx,dy,dz)=} {(rx,ry,rz)=}")
one_hot_gt: Tensor = class2one_hot(torch.tensor(resized_gt[None, ...],
dtype=torch.int64),
K=2)[0]
assert one_hot_gt.shape == (2, 256, 256, z), one_hot_gt.shape
distmap: np.ndarray = one_hot2dist(one_hot_gt.numpy(),
resolution=(rx, ry, rz),
dtype=np.float32)
save_dir_flair: Path = Path(dest_dir, flair_dir)
save_dir_t1: Path = Path(dest_dir, t1_dir)
save_dir_gt: Path = Path(dest_dir, gt_dir)
save_dir_in_npy: Path = Path(dest_dir, in_npy_dir)
save_dir_gt_npy: Path = Path(dest_dir, gt_npy_dir)
save_dir_distmap_npy: Path = Path(dest_dir, "3d_distmap")
save_dirs = [save_dir_flair, save_dir_t1, save_dir_gt]
for j in range(flair.shape[-1]):
flair_s = resized_flair[:, :, j]
t1_s = resized_t1[:, :, j]
gt_s = resized_gt[:, :, j]
dist_s = distmap[:, :, :, j]
# if gt_s.sum() > 0:
# print(f"{dist_s.min()=} {dist_s.max()=}")
# _, axes = plt.subplots(nrows=1, ncols=2)
# axes[0].imshow(gt_s)
# axes[0].set_title("GT")
# tmp = axes[1].imshow(dist_s[1, ...])
# axes[1].set_title("Distance map")
# plt.colorbar(tmp, ax=axes[1])
# plt.show()
slices = [flair_s, t1_s, gt_s]
assert flair_s.shape == t1_s.shape == gt_s.shape == dist_s[
0, ...].shape, ((x, y, z), flair_s.shape, dist_s.shape)
# gt_s[np.where(gt_s == 2)] = 0 # Now do that part earlier
assert set(np.unique(gt_s)).issubset([0, 1]), np.unique(gt_s)
if discard_negatives and (gt_s.sum() == 0):
continue
for k in range(n_augment + 1):
if k == 0:
to_save = slices
else:
to_save = map_(np.asarray, augment_arr(*slices))
assert to_save[0].shape == slices[0].shape, (to_save[0].shape,
slices[0].shape)
filename = f"{p_id}_{k:02d}_{j:04d}"
space_dict[filename] = (rx, ry)
for save_dir, data in zip(save_dirs, to_save):
save_dir.mkdir(parents=True, exist_ok=True)
if "gt" not in str(save_dir):
img = (data * 255).astype(np.uint8)
else:
img = data.astype(np.uint8)
with warnings.catch_warnings():
warnings.filterwarnings("ignore", category=UserWarning)
imsave(str(Path(save_dir, filename).with_suffix(".png")),
img)
multimodal = np.stack(
to_save[:-1]) # Do not include the ground truth
assert 0 <= multimodal.min(), multimodal.min()
assert multimodal.max() <= 1, multimodal.max()
save_dir_in_npy.mkdir(parents=True, exist_ok=True)
save_dir_gt_npy.mkdir(parents=True, exist_ok=True)
np.save(
Path(save_dir_in_npy, filename).with_suffix(".npy"),
multimodal)
np.save(
Path(save_dir_gt_npy, filename).with_suffix(".npy"),
to_save[-1])
save_dir_distmap_npy.mkdir(parents=True, exist_ok=True)
np.save(
Path(save_dir_distmap_npy, filename).with_suffix(".npy"),
dist_s)
return neg, pos, space_dict, (rx, ry, rz)