def main():
parser = argparse.ArgumentParser()
parser.add_argument('--yaml-config',
type=str,
default='simg_bmi_regression_0_cam.yaml')
args = parser.parse_args()
SRC_ROOT = os.path.dirname(os.path.realpath(__file__)) + '/../..'
yaml_config = os.path.join(SRC_ROOT, f'src/yaml/{args.yaml_config}')
logger.info(f'Read yaml file {yaml_config}')
f = open(yaml_config, 'r').read()
config = yaml.safe_load(f)
num_fold = config['fold_num']
exp_dir = config['exp_dir']
out_dir = osp.join(exp_dir, 'cam_analysis/atlas_space')
mkdir_p(out_dir)
file_list = read_file_contents_list(file_list_txt)
# folder_obj = DataFolder(ori_img_dir, file_list)
folder_obj = DataFolder(ori_atlas_dir, file_list)
ave_obj = AverageValidRegion(folder_obj, 10)
ave_obj.run_get_average()
ave_obj.output_result_folder(out_dir, ambient_val=-1000)
def main():
file_list = read_file_contents_list(in_file_list_file)
label_reader_obj = ClinicalDataReaderSPORE.create_spore_data_reader_csv(
in_label_file)
bmi_list, valid_file_list = label_reader_obj.get_gt_value_BMI(file_list)
# print(bmi_list[:10])
train_idx_fold_array, valid_idx_fold_array, test_idx_fold_array = get_idx_list_array_bmi_session_level_split(
bmi_list, num_fold)
test_bmi_list_array = []
for idx_fold in range(num_fold):
test_bmi_list = [
bmi_list[data_idx] for data_idx in test_idx_fold_array[idx_fold]
]
test_bmi_list_array.append(test_bmi_list)
out_png = os.path.join(out_folder, 'fold_distribution.png')
fig, ax = plt.subplots(figsize=(10, 7))
ax.hist(test_bmi_list_array,
bins=5,
label=[f'fold-{idx}' for idx in range(5)],
alpha=0.7,
rwidth=0.8,
density=False)
ax.legend(loc='best')
plt.grid(axis='y', alpha=0.8)
plt.xlabel('BMI')
# plt.ylabel('Density (Count / Total)')
plt.ylabel('Count')
print(f'Save image to {out_png}')
plt.savefig(out_png, bbox_inches='tight', pad_inches=0.1)
plt.close()
def get_statics_subgroup(file_list_txt, in_csv):
# out_csv = os.path.join(exp_dir, 'pred_total.csv')
print(f'Load csv {in_csv}')
data_df = pd.read_csv(in_csv, index_col='file_name')
study_file_list = read_file_contents_list(file_list_txt)
num_study_case = len(study_file_list)
print(f'# Study files: {num_study_case}')
study_data_df = data_df.loc[study_file_list]
study_label_list = study_data_df['label'].to_numpy()
study_pred1_list = study_data_df['pred1'].to_numpy()
study_pred2_list = study_data_df['pred2'].to_numpy()
diff1 = study_pred1_list - study_label_list
diff2 = study_pred2_list - study_label_list
diff1_rmse = np.sqrt(np.mean(np.power(np.abs(diff1), 2)))
diff1_mean = np.mean(diff1)
diff2_rmse = np.sqrt(np.mean(np.power(np.abs(diff2), 2)))
diff2_mean = np.mean(diff2)
print(f'Diff1 RMSE: {diff1_rmse:.3f}')
print(f'Diff1 Mean: {diff1_mean:.3f}')
print(f'Diff2 RMSE: {diff2_rmse:.3f}')
print(f'Diff2 Mean: {diff2_mean:.3f}')
def axial_clip_plot_native():
file_name_list = read_file_contents_list(file_list_txt)
for file_name in file_name_list:
in_img_path = os.path.join(in_folder, file_name)
cliper_obj = ClipPlotSeriesWithBack(in_img_path, None, None, 10, 35,
15, 1, -255, 255, None, None, None)
cliper_obj.clip_plot_img_only(out_png_folder)
def plot_round_normal_distribution():
file_list_round = read_file_contents_list(out_file_list_round_fov)
file_list_normal = read_file_contents_list(out_file_list_normal_fov)
label_df = pd.read_csv(in_csv_file, index_col='id')
file_list_round_no_ext = [
file_name.replace('.nii.gz', '') for file_name in file_list_round
]
file_list_normal_no_ext = [
file_name.replace('.nii.gz', '') for file_name in file_list_normal
]
bmi_round = label_df.loc[file_list_round_no_ext]['bmi'].to_numpy()
bmi_normal = label_df.loc[file_list_normal_no_ext]['bmi'].to_numpy()
data_array_sequence = []
data_array_sequence.append(bmi_round)
data_array_sequence.append(bmi_normal)
f, ax = plt.subplots(figsize=(8, 6))
ax.hist(data_array_sequence,
bins=10,
color=['r', 'b'],
label=[
f'Round FOV ({len(bmi_round)})',
f'Normal FOV ({len(bmi_normal)})'
],
alpha=0.8,
rwidth=0.9,
density=True)
ax.legend(loc='best')
plt.grid(axis='y', alpha=0.8)
plt.xlabel('BMI')
plt.ylabel('Density (Count / Total)')
print(f'Save image to {out_fov_type_bmi_hist_plot}')
plt.savefig(out_fov_type_bmi_hist_plot,
bbox_inches='tight',
pad_inches=0.1)
plt.close()
def plot_round_fov_output(yaml_config):
"""
Get the output plot on round fov cases. Run regression_agreement_analysis before this.
:param yaml_config:
:return:
"""
SRC_ROOT = os.path.dirname(os.path.realpath(__file__)) + '/../..'
yaml_config = os.path.join(SRC_ROOT, f'src/yaml/{yaml_config}')
logger.info(f'Read yaml file {yaml_config}')
f = open(yaml_config, 'r').read()
config = yaml.safe_load(f)
# num_fold = config['fold_num']
exp_dir = config['exp_dir']
in_csv = os.path.join(exp_dir, 'pred_total.csv')
data_df = pd.read_csv(in_csv, index_col='file_name')
round_fov_file_list = read_file_contents_list(out_file_list_round_fov)
_get_rmse_mean_shift_group(round_fov_file_list, data_df)
normal_fov_file_list = read_file_contents_list(out_file_list_normal_fov)
_get_rmse_mean_shift_group(normal_fov_file_list, data_df)
all_file_list = read_file_contents_list(file_list_all)
_get_rmse_mean_shift_group(all_file_list, data_df)
round_fov_df = data_df.loc[round_fov_file_list]
round_fov_pred_list = round_fov_df['pred'].to_numpy()
round_fov_label_list = round_fov_df['label'].to_numpy()
out_mean_diff_png = os.path.join(exp_dir, 'round_fov_mean_diff.png')
out_scatter_png = os.path.join(exp_dir, 'round_fov_scatter.png')
mean_diff_plot(round_fov_pred_list, round_fov_label_list, np.array([0]),
out_mean_diff_png)
scatter_plot(round_fov_pred_list, round_fov_label_list, np.array([0]),
out_scatter_png)
out_pred_csv = os.path.join(exp_dir, 'pred_total_round_fov.csv')
round_fov_df.to_csv(out_pred_csv)
def main():
parser = argparse.ArgumentParser()
parser.add_argument('--yaml-config',
type=str,
default='simg_bmi_regression_0_cam.yaml')
args = parser.parse_args()
file_list = read_file_contents_list(file_list_txt)
# folder_obj = DataFolder(ori_img_dir, file_list)
folder_obj = DataFolder(affine_img_dir, file_list)
ave_obj = AverageValidRegion(folder_obj, 10)
ave_obj.run_get_average()
ave_obj.output_result_folder(out_folder, ambient_val=-1000)
def get_list_round_and_normal_fov():
file_name_list = read_file_contents_list(file_list_txt)
round_list = []
normal_list = []
for file_name in file_name_list:
img_obj = ScanWrapper(os.path.join(in_native_folder, file_name))
img_data = img_obj.get_data()
img_thres_data = (img_data < threshold_val).astype(int)
if np.sum(img_thres_data) > thres_num_voxel:
print(f'Round FOV: {file_name}')
round_list.append(file_name)
else:
print(f'Normal FOV: {file_name}')
normal_list.append(file_name)
print(f'# Total round FOV: {len(round_list)}')
print(f'# Total normal FOV: {len(normal_list)}')
save_file_contents_list(out_file_list_round_fov, round_list)
save_file_contents_list(out_file_list_normal_fov, normal_list)
def get_missing_roi_ratio():
file_name_list = read_file_contents_list(file_list_txt)
data_dict = []
num_total = len(file_name_list)
counter = 0
for file_name in file_name_list:
logger.info(f'Analyze {file_name} ({counter} / {num_total})')
missing_ratio = _get_sess_missing_ratio(file_name)
logger.info(f'missing ratio - {missing_ratio:.3f}')
counter += 1
data_dict.append({
'file_name': file_name,
'missing_ratio': missing_ratio
})
missing_ratio_df = pd.DataFrame(data_dict)
missing_ratio_df.to_csv(out_missing_ratio_csv)