def _run_single_scan(self, idx):
in_img = ScanWrapper(self._in_data_folder.get_file_path(idx))
in_mask = ScanWrapper(self._in_mask_folder_obj.get_file_path(idx))
in_img_data = in_img.get_data()
in_mask_data = in_mask.get_data()
mask_data = np.zeros(in_mask_data.shape, dtype=int)
for lung_label in self._lung_label_list:
mask_match_map = (in_mask_data == lung_label).astype(int)
print(f'lung label: {lung_label}')
print(f'match_map shape: {mask_match_map.shape}')
print(f'num posi voxels: {np.sum(mask_match_map)}')
mask_data += mask_match_map
mask_data = ~mask_data.astype('bool')
print(f'Final mask: {np.sum(mask_data)}')
in_img_data_with_mask = np.ma.array(in_img_data, mask=mask_data.astype('bool'))
mean_in_mask = in_img_data_with_mask.mean()
file_name = self._in_data_folder.get_file_name(idx)
result_dict = {
'file_name': file_name,
'mean': mean_in_mask
}
print(result_dict)
return result_dict
def _run_single_scan(self, idx):
in_file_1_path = self._in_data_folder.get_file_path(idx)
in_file_2_path = self._in_data_folder_2.get_file_path(idx)
in_img_1 = ScanWrapper(in_file_1_path).get_data()
in_img_2 = ScanWrapper(in_file_2_path).get_data()
nrmse = compare_nrmse(np.abs(in_img_1), np.abs(in_img_2))
self._nrmse_diff.append(nrmse)
def clip_plot_3_view(self, out_png_folder):
in_img_obj = ScanWrapper(self._in_img_path)
in_mask_obj = ScanWrapper(self._in_mask_path)
in_img_data = in_img_obj.get_data()
in_mask_data = in_mask_obj.get_data()
masked_img_data = np.zeros(in_img_data.shape, dtype=float)
masked_img_data.fill(np.nan)
masked_img_data[in_mask_data == 1] = in_img_data[in_mask_data == 1]
def clip_plot(self, out_png_folder):
in_img_obj = ScanWrapper(self._in_img_path)
in_back_obj = ScanWrapper(self._in_back_img_path)
in_img_data = in_img_obj.get_data()
in_back_data = in_back_obj.get_data()
masked_img_data = None
masked_back_data = None
if self._in_mask_path is not None:
in_mask_obj = ScanWrapper(self._in_mask_path)
in_mask_data = in_mask_obj.get_data()
masked_img_data = np.zeros(in_img_data.shape, dtype=float)
masked_img_data.fill(np.nan)
masked_img_data[in_mask_data == 1] = in_img_data[in_mask_data == 1]
masked_back_data = np.zeros(in_back_data.shape, dtype=float)
masked_back_data.fill(np.nan)
masked_back_data[in_mask_data == 1] = in_back_data[in_mask_data == 1]
else:
masked_img_data = in_img_data
masked_back_data = in_back_data
self._plot_stacked_view(
self._num_clip,
self._step_axial,
masked_img_data,
masked_back_data,
'axial',
out_png_folder,
1
)
self._plot_stacked_view(
self._num_clip,
self._step_sagittal,
masked_img_data,
masked_back_data,
'sagittal',
out_png_folder,
5.23438 / 2.28335
)
self._plot_stacked_view(
self._num_clip,
self._step_coronal,
masked_img_data,
masked_back_data,
'coronal',
out_png_folder,
5.23438 / 2.17388
)
def _run_single_scan(self, idx):
in_ori_data = ScanWrapper(
self._in_data_folder.get_file_path(idx)).get_data()
in_jac_data = ScanWrapper(
self._in_jac_folder.get_file_path(idx)).get_data()
self._data_matrix[self._cur_idx, :self._ref_ori.get_number_voxel(
)] = convert_3d_2_flat(in_ori_data)
self._data_matrix[
self._cur_idx,
self._ref_ori.get_number_voxel():] = convert_3d_2_flat(in_jac_data)
self._cur_idx += 1
def __init__(self,
config,
in_ori_folder,
in_jac_folder,
batch_size,
file_list_txt=None):
super().__init__(config, in_ori_folder, file_list_txt)
self._in_jac_folder = DataFolder(in_jac_folder, file_list_txt)
self._ref_ori = ScanWrapper(self._in_data_folder.get_file_path(0))
self._ref_jac = ScanWrapper(self._in_jac_folder.get_file_path(0))
self._chunk_list = self._in_data_folder.get_chunks_list_batch_size(
batch_size)
self._data_matrix = []
self._cur_idx = 0
def get_average_map(file_list, save_path):
first_img = ScanWrapper(file_list[0])
im_shape = first_img.get_shape()
sum_map = np.zeros(im_shape, dtype=float)
for idx_image in range(len(file_list)):
img = ScanWrapper(file_list[idx_image])
img_data = img.get_data()
print(f'Adding {file_list[idx_image]} ({idx_image} / {len(file_list)})')
print(f'Max intensity {np.max(img_data)}')
sum_map += img_data
average_map = sum_map / float(len(file_list))
print(f'Average map max int {np.max(average_map)}')
first_img.save_scan_same_space(save_path, average_map)
def _run_single_scan(self, idx):
in_file_path = self._in_data_folder.get_file_path(idx)
in_data = ScanWrapper(in_file_path)
in_img = in_data.get_data()
# self._data_matrix[idx, :] = in_img.reshape(in_data.get_number_voxel())
self._data_matrix[idx, :] = convert_3d_2_flat(in_img)
def __init__(self, config, in_folder, mask_img, out_png_folder, file_list_txt):
super().__init__(config, in_folder, file_list_txt)
self._mask_img = ScanWrapper(mask_img)
self._out_png_folder = DataFolder.get_data_folder_obj(config, out_png_folder, data_list_txt=file_list_txt)
self._out_png_folder.change_suffix('.png')
self._vmax = 500
self._vmin = -1000
def plot_pc(self, out_png):
fig = plt.figure(figsize=(self._num_show_pc * 20,
self._num_clip * self._num_view * 15))
gs = gridspec.GridSpec(1, self._num_show_pc)
sub_gs_list = []
for idx_pc in range(self._num_show_pc):
sub_gs = gs[idx_pc].subgridspec(self._num_clip * self._num_view, 1)
sub_gs_list.append(sub_gs)
for idx_pc in range(self._num_show_pc):
img_data_obj = ScanWrapper(
self._pc_folder_obj.get_file_path(idx_pc))
img_data = img_data_obj.get_data()
img_name = os.path.basename(
self._pc_folder_obj.get_file_path(idx_pc)).replace(
'.nii.gz', '')
logger.info(
f'Reading image {self._pc_folder_obj.get_file_path(idx_pc)}')
self._plot_one_pc(img_data, sub_gs_list[idx_pc], f'{img_name}')
# out_eps = out_png.replace('.png', '.eps')
logger.info(f'Save fig to {out_png}')
# plt.savefig(out_eps, bbox_inches='tight', pad_inches=0, dpi=self._out_dpi, format='pdf')
plt.savefig(out_png,
bbox_inches='tight',
pad_inches=0,
dpi=self._out_dpi)
plt.close(fig=fig)
def main():
parser = argparse.ArgumentParser(description='Load a saved pca object')
parser.add_argument('--load-pca-bin-path', type=str,
help='Location of the pca bin')
parser.add_argument('--in-folder', type=str)
parser.add_argument('--out-folder', type=str)
parser.add_argument('--mask-img', type=str)
parser.add_argument('--num-mode', type=int)
parser.add_argument('--file-list-txt', type=str)
parser.add_argument('--num-process', type=int, default=10)
args = parser.parse_args()
in_folder_obj = DataFolder(args.in_folder, args.file_list_txt)
out_folder_obj = DataFolder(args.out_folder, args.file_list_txt)
mask_img_obj = ScanWrapper(args.mask_img)
paral_obj = ParalGetResidual(
args.load_pca_bin_path,
in_folder_obj,
out_folder_obj,
mask_img_obj,
args.num_mode,
args.num_process
)
paral_obj.run_parallel()
def __init__(self, config, in_folder, ref_img, batch_size, file_list=None):
super().__init__(config, in_folder, file_list)
self._ref_img = ScanWrapper(self._in_data_folder.get_file_path(0))
self._chunk_list = self._in_data_folder.get_chunks_list_batch_size(
batch_size)
self._data_matrix = []
self._cur_idx = 0
def __init__(self, config, in_dat_folder, out_folder, ref_img,
file_list_txt):
super().__init__(config, in_dat_folder, file_list_txt)
self._ref_img = ScanWrapper(ref_img)
self._out_data_folder = DataFolder.get_data_folder_obj(
config, out_folder, data_list_txt=file_list_txt)
self._jacobian_tool = config['deedsBCV_jacobian']
def _run_chunk(self, chunk_list):
pca_nii_3d = PCA_NII_3D(None, None, 1)
pca_nii_3d.load_pca(self._pca_bin_path)
result_list = []
for idx in chunk_list:
self._in_data_folder.print_idx(idx)
image_obj = ScanWrapper(self._in_data_folder.get_file_path(idx))
jac_obj = ScanWrapper(self._in_jac_folder_obj.get_file_path(idx))
low_dim_representation = pca_nii_3d.transform_concat(image_obj, jac_obj)
result = {
'scan_name': self._in_data_folder.get_file_name(idx),
'low_dim': low_dim_representation
}
print(result)
result_list.append(result)
return result_list
def __init__(self, config, in_folder, out_folder, ref_img, order=3):
super().__init__(config, in_folder)
self._out_data_folder = DataFolder.get_data_folder_obj(
config, out_folder)
self._ref_img = ScanWrapper(ref_img)
self._order = order
self._reg_resample = config['niftyreg_resample']
self._reg_resmaple_ref = config['reg_resample_ref_img']
def __init__(self, config, in_jac_folder, in_affine_mat_folder, out_folder,
ref_img, file_list_txt):
super().__init__(config, in_jac_folder, file_list_txt)
self._ref_img = ScanWrapper(ref_img)
self._in_affine_mat_folder = DataFolder.get_data_folder_obj(
config, in_affine_mat_folder, data_list_txt=file_list_txt)
self._out_data_folder = DataFolder.get_data_folder_obj(
config, out_folder, data_list_txt=file_list_txt)
def _run_single_scan(self, idx):
in_img = ScanWrapper(self._in_data_folder.get_file_path(idx))
out_mask_path = self._out_folder.get_file_path(idx)
in_img_data = in_img.get_data()
non_nan_mask = in_img_data == in_img_data
logger.info(f'Save non-nan mask to {out_mask_path}')
in_img.save_scan_same_space(out_mask_path, non_nan_mask.astype(int))
def clip_plot(self, out_png_folder):
in_int_obj = ScanWrapper(self._in_int_path)
in_jac_obj = ScanWrapper(self._in_jac_path)
in_att_obj = ScanWrapper(self._in_att_path)
in_int_data = in_int_obj.get_data()
in_jac_data = in_jac_obj.get_data()
in_att_data = in_att_obj.get_data()
fig, axs = plt.subplots(1, 3, constrained_layout=True, figsize=(30, 30))
# plt.axis('off')
for ax in axs:
ax.axis('off')
self._plot_view(
self._num_clip,
self._step_axial,
in_int_data, in_jac_data, in_att_data,
'axial', 1, axs[0]
)
self._plot_view(
self._num_clip,
self._step_coronal,
in_int_data, in_jac_data, in_att_data,
'coronal', 5.23438 / 2.17388, axs[1]
)
self._plot_view(
self._num_clip,
self._step_sagittal,
in_int_data, in_jac_data, in_att_data,
'sagittal', 5.23438 / 2.28335, axs[2]
)
out_root_folder = os.path.join(out_png_folder, 'stacked')
mkdir_p(out_root_folder)
out_png_path = os.path.join(out_root_folder, f'{self._in_img_file_name}.png')
print(f'Save overlay png to {out_png_path}')
plt.savefig(out_png_path, bbox_inches='tight', pad_inches=0)
plt.close()
def _run_single_scan(self, idx):
in_file_path = self._in_data_folder.get_file_path(idx)
out_file_path = self._out_data_folder.get_file_path(idx)
in_img = ScanWrapper(in_file_path).get_data()
average_img = self._average_img.get_data()
np.copyto(in_img, average_img, where=(in_img != in_img))
np.copyto(in_img, 0, where=(in_img != in_img))
self._average_img.save_scan_same_space(out_file_path, in_img)
def _run_chunk(self, chunk_list):
result_list = []
im_shape = self._ref_img.get_shape()
sum_image_union = np.zeros(im_shape)
region_mask_count_image = np.zeros(im_shape)
for idx in chunk_list:
self._in_data_folder.print_idx(idx)
img_obj = ScanWrapper(self._in_data_folder.get_file_path(idx))
mask_obj = ScanWrapper(
self._in_region_mask_folder.get_file_path(idx))
sum_image_union += img_obj.get_data()
region_mask_count_image += mask_obj.get_data()
result = {
'sum_image': sum_image_union,
'region_count': region_mask_count_image
}
result_list.append(result)
return result_list
def __init__(self,
config,
in_folder,
out_folder,
average_img,
file_list_txt=None):
super().__init__(config, in_folder, file_list_txt=file_list_txt)
self._out_data_folder = DataFolder.get_data_folder_obj(
config, out_folder, data_list_txt=file_list_txt)
mkdir_p(out_folder)
self._average_img = ScanWrapper(average_img)
def __init__(self,
config,
in_folder=None,
data_file_txt=None,
in_data_folder_obj=None):
self._data_folder = None
if in_data_folder_obj is None:
self._data_folder = DataFolder(in_folder, data_file_txt)
else:
self._data_folder = in_data_folder_obj
self._standard_ref = ScanWrapper(self._data_folder.get_first_path())
self._num_processes = config['num_processes']
def _plot_pc_one_view(self, idx_view, out_png_path):
view_config_list = self._get_view_config()
clip_plane = view_config_list[idx_view]['clip plane']
num_files = self._pc_folder_obj.num_files()
num_lines = int(num_files / self._num_show_pc)
fig = plt.figure(figsize=(self._num_show_pc * 15, num_lines * 12))
gs = gridspec.GridSpec(num_lines, self._num_show_pc)
gs.update(wspace=0.025, hspace=0.025)
for idx_row in range(num_lines):
# if idx_row != 0:
# continue
for idx_column in range(self._num_show_pc):
# if idx_column != 0:
# continue
idx_file = idx_column + self._num_show_pc * idx_row
logger.info(
f'Plot scan {self._pc_folder_obj.get_file_path(idx_file)}')
img_data_obj = ScanWrapper(
self._pc_folder_obj.get_file_path(idx_file))
img_data = img_data_obj.get_data()
range_norm = np.max([np.min(img_data), np.max(img_data)])
# vmin = - 0.5 * range_norm
# vmax = 0.5 * range_norm
vmin = -0.0015
vmax = 0.0015
clip = self._clip_image(img_data, clip_plane, 0)
ax = plt.subplot(gs[idx_row, idx_column])
plt.axis('off')
im = plt.imshow(clip,
interpolation='none',
cmap=self._cm,
norm=colors.Normalize(vmin=vmin, vmax=vmax))
# divider = make_axes_locatable(ax)
# cax = divider.append_axes("right", size="5%", pad=0.05)
#
# cb = plt.colorbar(im, cax=cax)
# # cb.set_label('Intensity of eigen image')
# cb.ax.tick_params(labelsize=self._sub_title_font_size / 2)
logger.info(f'Save plot to {out_png_path}')
plt.savefig(out_png_path,
bbox_inches='tight',
pad_inches=0.1,
dpi=self._out_dpi)
plt.close(fig=fig)
def __init__(self,
config,
in_scan_folder,
in_region_mask_folder,
file_list_txt,
out_average_img,
ambient=-5000):
super().__init__(config, in_scan_folder, file_list_txt)
self._in_region_mask_folder = DataFolder.get_data_folder_obj(
config, in_region_mask_folder, data_list_txt=file_list_txt)
self._out_img_path = out_average_img
self._ref_img = ScanWrapper(self._in_data_folder.get_first_path())
self._ambient = ambient
def _run_single_scan(self, idx):
in_data_path = self._in_data_folder.get_file_path(idx)
in_affine_mat_path = self._in_affine_mat_folder.get_file_path(
idx).replace('.nii.gz', '.txt')
out_file_path = self._out_data_folder.get_file_path(idx)
in_image = ScanWrapper(in_data_path)
affine_mat = self._get_affine_matrix(in_affine_mat_path)
new_log_jacobian_det = in_image.get_data() + np.log(
np.linalg.det(affine_mat))
self._ref_img.save_scan_same_space(out_file_path, new_log_jacobian_det)
def __init__(self,
mask_img_path,
config,
in_ori_folder,
in_jac_folder,
batch_size,
file_list_txt=None):
super().__init__(config, in_ori_folder, in_jac_folder, batch_size,
file_list_txt)
self._mask_img = ScanWrapper(mask_img_path)
self._masked_ref_ori = ScanWrapperWithMask(
self._in_data_folder.get_file_path(0), self._mask_img.get_path())
self._masked_ref_jac = ScanWrapperWithMask(
self._in_jac_folder.get_file_path(0), self._mask_img.get_path())
def _get_slices_dim_list(pc_folder, file_list_txt):
center_slices_x = []
center_slices_y = []
center_slices_z = []
data_folder = DataFolder(pc_folder, file_list_txt)
for file_idx in range(show_pc_number):
data_folder.print_idx(file_idx)
file_path = data_folder.get_file_path(file_idx)
scan = ScanWrapper(file_path)
slice_x, slice_y, slice_z = scan.get_center_slices()
center_slices_x.append(slice_x)
center_slices_y.append(slice_y)
center_slices_z.append(slice_z)
return center_slices_x, center_slices_y, center_slices_z
def _run_single_scan(self, idx):
in_data_path = self._in_data_folder.get_file_path(idx)
in_img = ScanWrapper(in_data_path)
mse_val = self._calc_nmse(self._ref_img, in_img)
nmi_val = self._calc_nmi(self._niftyreg_reg_measure_path,
self._ref_img, in_img)
val = {
'Scan': self._in_data_folder.get_file_name(idx),
'MSE': mse_val,
'NMI': nmi_val
}
print(val)
return val
def _run_single_scan(self, idx):
in_img_path = self._in_data_folder.get_file_path(idx)
in_img = ScanWrapper(in_img_path)
slice_in_img = self._clip_nifti(in_img.get_data())
slice_mask_img = self._clip_nifti(self._mask_img.get_data())
plt.figure(figsize=(15, 15))
plt.axis('off')
clip_x_nii_rescale = exposure.rescale_intensity(slice_in_img, in_range=(self._vmin, self._vmax), out_range=(0, 1))
clip_x_nii_rgb = color.gray2rgb(clip_x_nii_rescale)
plt.imshow(clip_x_nii_rgb, alpha=0.8)
plt.imshow(slice_mask_img,
interpolation='none',
cmap='jet',
norm=colors.Normalize(vmin=0, vmax=1),
alpha=0.3)
out_png_path = self._out_png_folder.get_file_path(idx)
print(f'Saving image to {out_png_path}')
plt.savefig(out_png_path, bbox_inches='tight', pad_inches=0)
def __init__(self, config, in_folder, ref_img, file_list_txt):
super().__init__(config, in_folder, file_list_txt)
self._ref_img = ScanWrapper(ref_img)
self._df_mse_table = None
self._niftyreg_reg_measure_path = config['niftyreg_reg_measure']