def normalize_lung_CT(**kwargs):
mean_values = []
var_values = []
MIN_BOUND = -1000
MAX_BOUND = 400
Z_MAX, Y_MAX, X_MAX = kwargs['Z_MAX'], kwargs['Y_MAX'], kwargs['X_MAX']
vox_spacing = kwargs['vox_spacing']
utils.init_dims3D(Z_MAX, Y_MAX, X_MAX, vox_spacing)
luna_subset_path = kwargs['src']
luna_save_path = kwargs['dst']
file_list=glob(luna_subset_path + "/" + "*.mhd")
img_spacing = (vox_spacing, vox_spacing, vox_spacing)
for img_file in file_list:
itk_img = sitk.ReadImage(img_file)
(x_space, y_space, z_space) = itk_img.GetSpacing()
spacing_old = (z_space, y_space, x_space)
img_array = sitk.GetArrayFromImage(itk_img) # indexes are z,y,x (notice the ordering)
img, mu, var = utils.resample_volume(img_array, spacing_old, img_spacing, bounds=(MIN_BOUND, MAX_BOUND))
utils.save_updated_image(img, luna_save_path+os.path.basename(img_file), itk_img.GetOrigin(), img_spacing)
mean_values.append(mu)
var_values.append(var)
dataset_mean = np.mean(mean_values)
dataset_stddev = np.sqrt(np.mean(var_values))
return (dataset_mean, dataset_stddev)
def normalize_lung_mask(**kwargs):
Z_MAX, Y_MAX, X_MAX = kwargs['Z_MAX'], kwargs['Y_MAX'], kwargs['X_MAX']
vox_spacing = kwargs['vox_spacing']
utils.init_dims3D(Z_MAX, Y_MAX, X_MAX, vox_spacing)
luna_seg_lungs_path = kwargs['src']
luna_seg_lungs_save_path = kwargs['dst']
file_list=glob(os.path.join(luna_seg_lungs_path, "*.mhd"))
img_spacing = (vox_spacing, vox_spacing, vox_spacing)
for img_file in file_list:
itk_img = sitk.ReadImage(img_file)
(x_space, y_space, z_space) = itk_img.GetSpacing()
spacing_old = (z_space, y_space, x_space)
img_array = sitk.GetArrayFromImage(itk_img) # indexes are z,y,x (notice the ordering)
img, _, _ = utils.resample_volume(img_array, spacing_old, img_spacing)
img[img < 1] = 0
utils.save_updated_image(img, os.path.join(luna_seg_lungs_save_path, os.path.basename(img_file)),
itk_img.GetOrigin(), img_spacing)
def normalize_nodule_mask(**kwargs):
luna_path = kwargs['orig']
pixel_count = 0
mask_count = 0
annotations = kwargs['annotations']
Z_MAX, Y_MAX, X_MAX = kwargs['Z_MAX'], kwargs['Y_MAX'], kwargs['X_MAX']
shape_max = (Z_MAX, Y_MAX, X_MAX)
vox_spacing = kwargs['vox_spacing']
tables_path = kwargs['tables']
utils.init_dims3D(Z_MAX, Y_MAX, X_MAX, vox_spacing)
x_list, y_list, z_list = [], [], []
luna_normal_path = kwargs['src']
luna_mask_path = kwargs['dst']
def get_boundaries(origin, offsets, params):
diam, center = params
diam3 = np.array((diam, diam, diam))
diamu = diam + vox_spacing
diam3u = np.array((diamu, diamu, diamu))
v_center = np.rint((center - origin)/vox_spacing)
v_lower = np.rint((center - diam3 - origin)/vox_spacing)
v_upper = np.rint((center + diam3u - origin)/vox_spacing)
v_center -= offsets
v_lower -= offsets
v_upper -= offsets
x_list.append(v_upper[2])
y_list.append(v_upper[1])
z_list.append(v_upper[0])
x_list.append(v_lower[2])
y_list.append(v_lower[1])
z_list.append(v_lower[0])
return (v_lower, v_center, v_upper)
def l2_norm(pointA, pointB):
point = pointA - pointB
return np.sqrt(np.dot(point, point))
def get_filename(case):
for f in file_list:
if case in f:
return(f)
def update_mask(mask, CT, bounds):
v_lower, v_center, v_upper = bounds
z_min, y_min, x_min = v_lower
z_max, y_max, x_max = v_upper
pixel_count = 0
radius = np.rint((z_max - z_min + vox_spacing)/2)
ct_thresh = -1000
bit_count = 0
#print(v_center)
for z in range(z_min, z_max):
for y in range(y_min, y_max):
for x in range(x_min, x_max):
if l2_norm(np.array((z, y, x)), v_center) > radius:
break
if CT[z][y][x] > ct_thresh:
mask[z][y][x] = 4
pixel_count += 1
bit_count += 1
#assert bit_count != 0
if bit_count == 0:
print(v_center)
origin_dict = utils.npz_load(os.path.join(tables_path, "origin_table"))
offset_dict = utils.npz_load(os.path.join(tables_path, "offset_table"))
file_list=glob(luna_normal_path+"*.mhd")
df_node = pd.read_csv(annotations)
df_node["file"] = df_node["seriesuid"].apply(get_filename)
df_node = df_node.dropna()
img_spacing = (vox_spacing, vox_spacing, vox_spacing)
count = 0
for img_file in file_list:
mask_count = 0
mini_df = df_node[df_node["file"]==img_file] #get all nodules associate with file
if len(mini_df) == 0:
continue
mask = np.zeros(shape_max, dtype=np.int16)
series = os.path.basename(img_file)[0:-4]
origin = origin_dict[series]
if origin[1] > 0 and origin[2] > 0:
origin[1] = -origin[1]
origin[2] = -origin[2]
offsets = offset_dict[series]
itk_img = sitk.ReadImage(img_file)
img_array = sitk.GetArrayFromImage(itk_img)
for i in range(len(mini_df)):
node_x = mini_df["coordX"].values[i]
node_y = mini_df["coordY"].values[i]
node_z = mini_df["coordZ"].values[i]
diam = mini_df["diameter_mm"].values[i]
params = (diam, np.array((node_z, node_y, node_x)))
bounds = get_boundaries(origin, offsets, params)
_, v_center, _ = bounds
if np.min(v_center) < 0:
print("origin: {} offsets: {}\n params: {} v_center: {}".format(
origin, offsets, params, v_center))
continue
bounds = np.array(bounds).astype(np.int16)
update_mask(mask, img_array, bounds)
mask_count += 1
assert mask_count != 0
itk_mask_img = sitk.GetImageFromArray(mask, isVector=False)
itk_mask_img.SetSpacing(img_spacing)
itk_mask_img.SetOrigin(origin)
sitk.WriteImage(itk_mask_img, luna_mask_path+'/'+os.path.basename(img_file))