def main():
parser = argparse.ArgumentParser('')
parser.add_argument('--file-list-total', type=str, default=default_list_total)
parser.add_argument('--file-list-exclude', type=str, default=default_list_exclude)
parser.add_argument('--file-list-out', type=str, default=default_list_out)
args = parser.parse_args()
file_list_total = read_file_contents_list(args.file_list_total)
file_list_exclude = read_file_contents_list(args.file_list_exclude)
file_list_reduced = [file_name for file_name in file_list_total if file_name not in file_list_exclude]
save_file_contents_list(args.file_list_out, file_list_reduced)
def main():
# file_name = 'sub-S03174_ses-E06315_run-7_bp-chest_ct.nii'
# file_name = 'sub-S03591_ses-E07300_run-3_bp-chest_ct.nii'
# file_name = '00000001time20131205.nii.gz'
file_name_list_txt = '/nfs/masi/COVID19_public/BIMCV-COVID19/ct_cxr_pair/data/centaur_pilot.24.file_list'
file_name_list = read_file_contents_list(file_name_list_txt)
in_nii_folder = '/nfs/masi/COVID19_public/BIMCV-COVID19/ct_cxr_pair/data/ct'
out_mask_nii_folder = '/nfs/masi/COVID19_public/BIMCV-COVID19/ct_cxr_pair/data/centaur_pilot.24'
out_png_folder = '/nfs/masi/COVID19_public/BIMCV-COVID19/ct_cxr_pair/data/centaur_pilot.24.clip'
mkdir_p(out_png_folder)
for file_name in file_name_list:
in_nii = os.path.join(in_nii_folder, file_name + '.gz')
out_nii = os.path.join(out_mask_nii_folder, file_name)
# out_png = os.path.join(out_png_folder, file_name.replace('.nii', '.png'))
# get_pad_mask2(in_nii, out_nii)
# out_axial_png = os.path.join(out_png_folder, file_name.replace('.nii', '_axial.png'))
# multiple_clip_overlay_with_mask(in_nii, out_nii, out_axial_png, view_flag='axial')
#
# out_sagittal_png = os.path.join(out_png_folder, file_name.replace('.nii', '_sagittal.png'))
# multiple_clip_overlay_with_mask(in_nii, out_nii, out_sagittal_png, view_flag='sagittal')
#
# out_coronal_png = os.path.join(out_png_folder, file_name.replace('.nii', '_coronal.png'))
# multiple_clip_overlay_with_mask(in_nii, out_nii, out_coronal_png, view_flag='coronal')
out_folder = os.path.join(
'/nfs/masi/COVID19_public/BIMCV-COVID19/ct_cxr_pair/data/centaur_pilot.24.clip',
file_name)
clip_connected_component_bb_center_with_mask(in_nii, out_nii,
out_folder)
def main():
parser = argparse.ArgumentParser('Plot box and scatter data.')
parser.add_argument('--file-list-total', type=str)
parser.add_argument('--file-list-out', type=str)
parser.add_argument('--num-file-select', type=int)
args = parser.parse_args()
file_list_total = read_file_contents_list(args.file_list_total)
file_list_out = random.choices(file_list_total, k=args.num_file_select)
save_file_contents_list(args.file_list_out, file_list_out)
def main():
# file_name = 'sub-S03174_ses-E06315_run-7_bp-chest_ct.nii'
# file_name = 'sub-S03591_ses-E07300_run-3_bp-chest_ct.nii'
# file_name = '00000001time20131205.nii.gz'
# file_name_list_txt = '/nfs/masi/COVID19_public/BIMCV-COVID19/ct_cxr_pair/data/centaur_pilot.24.file_list'
file_name_list_txt = '/nfs/masi/COVID19_public/TCIA_COVID_19_CT/COVID-19-20_miccai/COVID-19-20_v2/ct_list.txt'
file_name_list = read_file_contents_list(file_name_list_txt)
# in_nii_folder = '/nfs/masi/COVID19_public/BIMCV-COVID19/ct_cxr_pair/data/ct'
# out_mask_nii_folder = '/nfs/masi/COVID19_public/BIMCV-COVID19/ct_cxr_pair/data/centaur_pilot.24'
# out_png_folder = '/nfs/masi/COVID19_public/BIMCV-COVID19/ct_cxr_pair/data/centaur_pilot.24.clip'
in_nii_folder = '/nfs/masi/COVID19_public/TCIA_COVID_19_CT/COVID-19-20_miccai/COVID-19-20_v2/Train'
out_mask_nii_folder = '/nfs/masi/COVID19_public/TCIA_COVID_19_CT/COVID-19-20_miccai/COVID-19-20_v2/Train'
out_png_folder = '/nfs/masi/COVID19_public/TCIA_COVID_19_CT/COVID-19-20_miccai/COVID-19-20_v2/clip_png'
mkdir_p(out_png_folder)
first_n = 20
counter = 0
for file_name in file_name_list:
if counter > first_n:
break
in_nii = os.path.join(in_nii_folder, file_name)
mask_nii = os.path.join(out_mask_nii_folder,
file_name.replace('_ct.', '_seg.'))
# out_png = os.path.join(out_png_folder, file_name.replace('.nii', '.png'))
# get_pad_mask2(in_nii, out_nii)
# out_axial_png = os.path.join(out_png_folder, file_name.replace('.nii', '_axial.png'))
# multiple_clip_overlay_with_mask(in_nii, out_nii, out_axial_png, view_flag='axial')
#
# out_sagittal_png = os.path.join(out_png_folder, file_name.replace('.nii', '_sagittal.png'))
# multiple_clip_overlay_with_mask(in_nii, out_nii, out_sagittal_png, view_flag='sagittal')
#
# out_coronal_png = os.path.join(out_png_folder, file_name.replace('.nii', '_coronal.png'))
# multiple_clip_overlay_with_mask(in_nii, out_nii, out_coronal_png, view_flag='coronal')
out_folder = os.path.join(out_png_folder, file_name)
clip_connected_component_bb_center_with_mask(in_nii, mask_nii,
out_folder)
counter += 1
def main():
parser = argparse.ArgumentParser('Plot box and scatter data.')
parser.add_argument('--file-list-folder', type=str)
parser.add_argument('--out-file-list', type=str)
args = parser.parse_args()
file_list_of_file_list_txt = os.listdir(args.file_list_folder)
total_list = []
for file_list_file in file_list_of_file_list_txt:
file_path = os.path.join(args.file_list_folder, file_list_file)
file_list = read_file_contents_list(file_path)
total_list = total_list + file_list
unique_list = unique(total_list)
logger.info(f'Number of unique files {len(unique_list)}')
with open(args.out_file_list, 'w') as file:
for item in unique_list:
file.write(item + '\n')
def main():
parser = argparse.ArgumentParser('Box plot for dice statistics.')
parser.add_argument('--dice-data-root', type=str)
parser.add_argument('--item-list', type=str)
parser.add_argument('--out-csv-folder', type=str)
args = parser.parse_args()
item_list = read_file_contents_list(args.item_list)
logger.info('Item list')
metric_list = get_metric_list(args.dice_data_root)
mean_table = []
std_table = []
for method_name in item_list:
metric_mean = {}
metric_std = {}
for metric in metric_list:
metric_dict = metric_list[metric]
csv_path = metric_dict['csv_root'] + '/' + method_name + '/' + metric_dict['csv_name']
mean, std = get_statics_csv(csv_path, metric_dict['column_name'])
metric_mean[metric] = mean
metric_std[metric] = std
metric_mean['Method'] = method_name
metric_std['Method'] = method_name
mean_table.append(metric_mean)
std_table.append(metric_std)
out_csv_mean = os.path.join(args.out_csv_folder, 'mean.csv')
out_csv_std = os.path.join(args.out_csv_folder, 'std.csv')
df_mean = pd.DataFrame(mean_table)
df_std = pd.DataFrame(std_table)
# headers = ["Method", "Dice_lung", "Dice_body", "MSD_lung", "MSD_body"]
pd.set_option('precision', 5)
headers = ["Method", "Dice_lung", "Dice_body"]
logger.info(f'Saving mean table to {out_csv_mean}')
df_mean.to_csv(out_csv_mean, index=False, columns=headers)
logger.info(f'Saving std table to {out_csv_std}')
df_std.to_csv(out_csv_std, index=False, columns=headers)
def main():
parser = argparse.ArgumentParser('Plot box and scatter data.')
parser.add_argument('--sub-file-list', type=str)
parser.add_argument('--in-csv', type=str)
parser.add_argument('--out-csv', type=str)
args = parser.parse_args()
data_df = pd.read_csv(args.in_csv)
scan_list_total = data_df['Scan'].tolist()
data_dict_total = data_df.to_dict('records')
sub_file_list = read_file_contents_list(args.sub_file_list)
idx_list = [
scan_list_total.index(file_name) for file_name in sub_file_list
]
sub_dict = [data_dict_total[idx] for idx in idx_list]
sub_data_df = pd.DataFrame(sub_dict)
logger.info(f'Save to {args.out_csv}')
sub_data_df.to_csv(args.out_csv, index=False)
def main():
parser = argparse.ArgumentParser('Box plot for dice statistics.')
parser.add_argument('--dice-data-root', type=str)
parser.add_argument('--item-list', type=str)
parser.add_argument('--out-fig-folder', type=str)
parser.add_argument('--num-scans', type=int, default=50)
args = parser.parse_args()
item_list = read_file_contents_list(args.item_list)
logger.info('Item list')
# print(item_list)
mask_flag_list = ['lung_mask', 'body_mask']
for mask_flag in mask_flag_list:
data_table = get_dice_data_table(args.dice_data_root, item_list,
args.num_scans, mask_flag)
mask_df = pd.DataFrame(data_table, columns=item_list)
mask_df.plot.box(grid='True')
out_fig_path = os.path.join(args.out_fig_folder, mask_flag + '.png')
logger.info(f'Save plot to {out_fig_path}')
plt.savefig(out_fig_path)
def _get_file_list(file_list_txt):
return read_file_contents_list(file_list_txt)
def main():
parser = argparse.ArgumentParser('Plot box and scatter data.')
parser.add_argument('--in-csv', type=str)
parser.add_argument('--thres-val', type=float)
parser.add_argument('--out-fig', type=str)
args = parser.parse_args()
test_collection_df = pd.read_csv(args.in_csv)
test_dict = test_collection_df.set_index('TestName').to_dict('index')
num_test = len(test_dict)
num_data = 50
y_table = np.full((num_data, num_test), 1).astype(float)
for test_idx, test_name in enumerate(test_dict):
test_item = test_dict[test_name]
test_df = pd.read_csv(test_item['CSV'])
y_all = test_df[test_item['COLUMN']].to_numpy()
y_table[:, test_idx] = y_all[:]
fig, ax = plt.subplots(figsize=(num_test * 2 + 2, 8))
plt.boxplot(y_table)
# Add outliers as scatter points
num_outlier = np.zeros(num_test)
mean_val = np.zeros(num_test)
# kp_val = np.zeros(num_test)
outlier_data = []
for test_idx, test_name in enumerate(test_dict):
test_item = test_dict[test_name]
test_df = pd.read_csv(test_item['CSV'])
data_dict = test_df.set_index('Scan').to_dict('index')
outlier_list = read_file_contents_list(test_item['OUTLIER'])
num_outlier[test_idx] = len(outlier_list)
scan_list = test_df['Scan'].to_list()
column_flag = test_item['COLUMN']
x_out_all, y_out_all = get_x_y_outlier_list(data_dict, column_flag,
scan_list, test_idx + 1)
# kp_val[test_idx] = data_dict[outlier_list[0]][column_flag]
mean_val[test_idx] = np.mean(y_out_all)
plt.scatter(x_out_all, y_out_all, color='r', alpha=0.5)
x_out, y_out = get_x_y_outlier_list(data_dict, column_flag,
outlier_list, test_idx + 1)
outlier_data.append(y_out)
# plot outlier as connected dots
# for outlier_idx in range(int(num_outlier[0])):
# y_val = [outlier_data[test_idx][outlier_idx] for test_idx in range(num_test)]
# plt.plot(range(1, len(y_val) + 1), y_val, linestyle='--', marker='o', color='b')
labels = [item.get_text() for item in ax.get_xticklabels()]
for test_idx, test_name in enumerate(test_dict):
mean = float("{:.5f}".format(mean_val[test_idx]))
# kp = float("{:.5f}".format(kp_val[test_idx]))
# labels[test_idx] = f'{test_name} \noutlier {int(num_outlier[test_idx])}/{num_data}\nmean {mean}'
labels[test_idx] = f'{test_name} \nmean {mean}'
# labels[test_idx] = f'{test_name}\n KP1 Lung DSC {kp}'
ax.set_xticklabels(labels)
# Threshold.
print(f'Thres: {args.thres_val}')
plt.axhline(y=args.thres_val, color='r', linestyle='--')
logger.info(f'Save plot to {args.out_fig}')
plt.grid()
plt.savefig(args.out_fig)
def main():
parser = argparse.ArgumentParser('Plot box and scatter data.')
parser.add_argument('--csv-data', type=str)
parser.add_argument('--column', type=str)
parser.add_argument('--out-fig', type=str)
parser.add_argument('--outlier-list-lung-dice', type=str)
parser.add_argument('--outlier-list-body-dice', type=str)
parser.add_argument('--outlier-list-nmi', type=str)
parser.add_argument('--outlier-list-manual', type=str)
parser.add_argument('--thres-val', type=float)
args = parser.parse_args()
logger.info(f'Read csv: {args.csv_data}')
data_df = pd.read_csv(args.csv_data)
data_dict = data_df.set_index('Scan').to_dict()[args.column]
# print(data_dict)
outlier_list_lung_dice = read_file_contents_list(
args.outlier_list_lung_dice)
outlier_list_body_dice = read_file_contents_list(
args.outlier_list_body_dice)
outlier_list_nmi = read_file_contents_list(args.outlier_list_nmi)
outlier_list_manual = read_file_contents_list(args.outlier_list_manual)
outlier_items = [{
'outlier_list': outlier_list_lung_dice,
'idx': 2,
'color': 'red'
}, {
'outlier_list': outlier_list_body_dice,
'idx': 3,
'color': 'blue'
}, {
'outlier_list': outlier_list_nmi,
'idx': 4,
'color': 'green'
}, {
'outlier_list': outlier_list_manual,
'idx': 5,
'color': 'orange'
}]
num_metric = 4
y_all = data_df[args.column].to_numpy()
y_all_table = np.zeros((len(y_all), num_metric + 1))
for i in range(num_metric + 1):
y_all_table[:, i] = y_all[:]
x_all = np.random.normal(1, 0.01, len(y_all))
fig, ax = plt.subplots()
plt.boxplot(y_all_table)
plt.scatter(x_all, y_all, c='grey', alpha=1)
for outlier_item in outlier_items:
x_outlier, y_outlier = get_x_y_outlier_list(
data_dict, outlier_item['outlier_list'], outlier_item['idx'])
plt.scatter(x_outlier, y_outlier, c=outlier_item['color'], alpha=0.5)
labels = [item.get_text() for item in ax.get_xticklabels()]
labels[0] = f'All ({len(y_all)})'
labels[1] = f'Outliers (Lung, {len(outlier_list_lung_dice)}/{len(y_all)})'
labels[2] = f'Outliers (Body, {len(outlier_list_body_dice)}/{len(y_all)})'
labels[3] = f'Outliers (NMI, {len(outlier_list_nmi)}/{len(y_all)})'
labels[
4] = f'Outliers (Manual QA, {len(outlier_list_manual)}/{len(y_all)})'
ax.set_xticklabels(labels)
# Threshold.
print(f'Thres: {args.thres_val}')
plt.axhline(y=args.thres_val, color='r', linestyle='--')
logger.info(f'Save plot to {args.out_fig}')
fig.set_size_inches(14, 7.5)
plt.savefig(args.out_fig)
def main():
parser = argparse.ArgumentParser('Plot box and scatter data.')
parser.add_argument('--csv-data-baseline', type=str)
parser.add_argument('--csv-data-1', type=str)
parser.add_argument('--csv-data-2', type=str)
parser.add_argument('--csv-data-3', type=str)
parser.add_argument('--csv-data-4', type=str)
parser.add_argument('--csv-data-5', type=str)
parser.add_argument('--outlier-list-baseline', type=str)
parser.add_argument('--outlier-list-1', type=str)
parser.add_argument('--outlier-list-2', type=str)
parser.add_argument('--outlier-list-3', type=str)
parser.add_argument('--outlier-list-4', type=str)
parser.add_argument('--outlier-list-5', type=str)
parser.add_argument('--column', type=str)
parser.add_argument('--thres-val', type=float)
parser.add_argument('--out-fig', type=str)
parser.add_argument('--num-complete-test', type=int)
args = parser.parse_args()
test_items = [
{
'in_csv': args.csv_data_baseline,
'outlier_list': args.outlier_list_baseline,
'name': 'Baseline',
'idx': 1
}, {
'in_csv': args.csv_data_1,
'outlier_list': args.outlier_list_1,
'name': 'Cap. Range 80mm',
'idx': 2
}
# },
# {
# 'in_csv': args.csv_data_2,
# 'outlier_list': args.outlier_list_2,
# 'name': 'Template ref',
# 'idx': 3
# },
# {
# 'in_csv': args.csv_data_3,
# 'outlier_list': args.outlier_list_3,
# 'name': '6 Step',
# 'idx': 4
# },
# {
# 'in_csv': args.csv_data_4,
# 'outlier_list': args.outlier_list_4,
# 'name': '4 Step',
# 'idx': 5
# },
# {
# 'in_csv': args.csv_data_5,
# 'outlier_list': args.outlier_list_5,
# 'name': 'Inst. Opt. (MH)',
# 'idx': 6
# }
]
num_test = len(test_items)
num_test_complete = args.num_complete_test
num_data = 50
y_table = np.full((num_data, num_test), 1).astype(float)
for test_idx in range(num_test_complete):
test_item = test_items[test_idx]
test_df = pd.read_csv(test_item['in_csv'])
y_all = test_df[args.column].to_numpy()
y_table[:, test_idx] = y_all[:]
fig, ax = plt.subplots()
plt.boxplot(y_table)
# Add outliers as scatter points
num_outlier = np.zeros(num_test)
mean_val = np.zeros(num_test)
for test_idx in range(num_test_complete):
test_item = test_items[test_idx]
test_df = pd.read_csv(test_item['in_csv'])
data_dict = test_df.set_index('Scan').to_dict()[args.column]
outlier_list = read_file_contents_list(test_item['outlier_list'])
num_outlier[test_idx] = len(outlier_list)
scan_list = test_df['Scan'].to_list()
x_out, y_out = get_x_y_outlier_list(data_dict, scan_list,
test_item['idx'])
mean_val[test_idx] = np.mean(y_out)
plt.scatter(x_out, y_out, color='r', alpha=0.5)
labels = [item.get_text() for item in ax.get_xticklabels()]
for test_idx in range(num_test):
test_item = test_items[test_idx]
test_name = test_item['name']
mean = float("{:.5f}".format(mean_val[test_idx]))
labels[
test_idx] = f'{test_name} \noutlier {int(num_outlier[test_idx])}/{num_data}\nmean {mean}'
ax.set_xticklabels(labels)
# Threshold.
print(f'Thres: {args.thres_val}')
plt.axhline(y=args.thres_val, color='r', linestyle='--')
logger.info(f'Save plot to {args.out_fig}')
plt.grid()
fig.set_size_inches(14, 7.5)
plt.savefig(args.out_fig)
def main():
parser = argparse.ArgumentParser('Plot box and scatter data.')
parser.add_argument('--csv-data', type=str)
parser.add_argument('--outlier-list-1', type=str)
parser.add_argument('--outlier-list-2', type=str)
parser.add_argument('--thres-val', type=float)
parser.add_argument('--out-fig', type=str)
args = parser.parse_args()
test_items = [{
'outlier_list': args.outlier_list_1,
'name': 'Diap Dist (mm)',
'idx': 1,
'title': 'Diaphragm Dist (mm)'
}, {
'outlier_list': args.outlier_list_2,
'name': 'Body Dist (mm)',
'idx': 2,
'title': 'Body Dist (mm)'
}]
num_data = 50
num_metric = 2
y_table = np.full((num_data, num_metric), 1).astype(float)
data_df = pd.read_csv(args.csv_data)
y_table[:, 0] = data_df['Diap Dist (mm)'].to_numpy()
y_table[:, 1] = data_df['Body Dist (mm)'].to_numpy()
fig, ax = plt.subplots()
plt.boxplot(y_table)
# Add outliers as scatter points
num_outlier = np.zeros(num_metric)
data_dict = data_df.set_index('Scan').to_dict()
for test_idx in range(len(test_items)):
test_item = test_items[test_idx]
data_dict_metric = data_dict[test_item['name']]
outlier_list = read_file_contents_list(test_item['outlier_list'])
num_outlier[test_idx] = len(outlier_list)
x_outlier, y_outlier = get_x_y_outlier_list(data_dict_metric,
outlier_list,
test_item['idx'])
scan_list_all = data_df['Scan'].to_list()
x_all, y_all = get_x_y_outlier_list(data_dict_metric, scan_list_all,
test_item['idx'])
plt.scatter(x_outlier, y_outlier, color='r', alpha=0.5)
plt.scatter(x_all, y_all, color='gray', alpha=0.3)
labels = [item.get_text() for item in ax.get_xticklabels()]
for test_idx in range(len(test_items)):
test_item = test_items[test_idx]
test_title = test_item['title']
labels[
test_idx] = f'{test_title} (Failed case {num_outlier[test_idx]}/{num_data})'
ax.set_xticklabels(labels)
# Threshold.
print(f'Thres: {args.thres_val}')
plt.axhline(y=args.thres_val, color='r', linestyle='--')
logger.info(f'Save plot to {args.out_fig}')
fig.set_size_inches(14, 7.5)
plt.savefig(args.out_fig)
