def create_image_patch_selection_process_log_image(case_id):
saliency_prediction_overview_visualization_directory_path = input_folder_path + "/visualizations/saliency_prediction_overview_visualization/" + case_id
most_salient_high_res_image_patch_directory_path = input_folder_path + "/visualizations/most_salient_image_patch_high_res/" + case_id
most_salient_high_res_image_patch_with_annotation_directory_path = input_folder_path + "/visualizations/highest_nuclei_count_annotations_on_most_salient_high_res_image_patch/" + case_id
image_patch_with_highest_nuclei_count_directory_path = input_folder_path + "/visualizations/image_patches_with_highest_nuclei_count/images_original/" + case_id
image_patch_with_highest_nuclei_count_directory_with_prediction_masks_path = input_folder_path + "/visualizations/image_patches_with_highest_nuclei_count/images_annotated/" + case_id
saliency_prediction_overview_visualization_path = path_utils.create_full_paths_to_files_in_directory_path(saliency_prediction_overview_visualization_directory_path)[0]
most_salient_high_res_image_patch_path = path_utils.create_full_paths_to_files_in_directory_path(most_salient_high_res_image_patch_directory_path)[0]
most_salient_high_res_image_patch_with_annotation_path = path_utils.create_full_paths_to_files_in_directory_path(most_salient_high_res_image_patch_with_annotation_directory_path)[0]
image_patch_with_highest_nuclei_count_path = path_utils.create_full_paths_to_files_in_directory_path(image_patch_with_highest_nuclei_count_directory_path)[0]
image_patch_with_highest_nuclei_count_with_prediction_masks_path = path_utils.create_full_paths_to_files_in_directory_path(image_patch_with_highest_nuclei_count_directory_with_prediction_masks_path)[0]
saliency_prediction_overview_visualization_image = Image.open(saliency_prediction_overview_visualization_path)
# most_salient_high_res_image_patch_image = Image.open(most_salient_high_res_image_patch_path)
# most_salient_high_res_image_patch_with_annotation_image = Image.open(
# most_salient_high_res_image_patch_with_annotation_path)
# image_patch_with_highest_nuclei_count_image = Image.open(image_patch_with_highest_nuclei_count_path)
# image_patch_with_highest_nuclei_count_with_prediction_masks_image = Image.open(
# image_patch_with_highest_nuclei_count_with_prediction_masks_path)
high_res_image_patch_images_merged_horizontally = image_utils.merge_images_horizontally(
[most_salient_high_res_image_patch_path, most_salient_high_res_image_patch_with_annotation_path])
image_patch_with_highest_nuclei_count_image_images_merged_horizontally = image_utils.merge_images_horizontally(
[image_patch_with_highest_nuclei_count_path,
image_patch_with_highest_nuclei_count_with_prediction_masks_path])
selection_processing_log_image = image_utils.merge_images_vertically(
[saliency_prediction_overview_visualization_image,
high_res_image_patch_images_merged_horizontally,
image_patch_with_highest_nuclei_count_image_images_merged_horizontally])
return selection_processing_log_image
def output_image_patch_with_highest_predicted_nuclei_count(
image_patches_path,
output_diretory_path,
all_image_patches_for_case_id_directory_path,
image_mode,
with_image_mask=False):
image_patch_paths = path_utils.create_full_paths_to_files_in_directory_path(
image_patches_path)
image_patch_path_with_highest_nuclei_count = get_image_patch_path_with_highest_nuclei_count(
image_patch_paths)
image_patch_file = image_patch_path_with_highest_nuclei_count.split(
'/')[-1]
if (with_image_mask is True):
new_image_patch_path = output_diretory_path + "/" + image_patch_file
shutil.copy(image_patch_path_with_highest_nuclei_count,
new_image_patch_path)
return
all_image_patches_mode_paths = path_utils.create_full_paths_to_directories_in_directory_path(
all_image_patches_for_case_id_directory_path)
directory_path_for_all_image_patches_with_specific_image_mode = get_path_for_all_image_patches_with_specific_image_mode(
all_image_patches_mode_paths, image_mode)
all_image_patches_with_specific_image_mode_paths = path_utils.create_full_paths_to_files_in_directory_path(
directory_path_for_all_image_patches_with_specific_image_mode)
parse_target_image_patch_file_name_to_dict = image_patch_file_name_parser.parse_image_patch_file_name_to_dict(
image_patch_file)
target_image_patch_metadata_object = image_patch_metadata_object.image_patch_metadata_object_from_image_patch_dict(
parse_target_image_patch_file_name_to_dict)
target_image_patch_path = get_target_image_patch_path(
all_image_patches_with_specific_image_mode_paths,
target_image_patch_metadata_object)
new_image_patch_path = output_diretory_path + "/" + image_patch_file
shutil.copy(target_image_patch_path, new_image_patch_path)
def output_image_patch_with_highest_predicted_nuclei_count(
image_patches_path, output_diretory_path):
image_patch_paths = path_utils.create_full_paths_to_files_in_directory_path(
image_patches_path)
image_patch_path_with_highest_nuclei_count = None
highest_nuclei_count = 0
for image_patch_path in image_patch_paths:
image_patch_file_name = image_patch_path.split('/')[-1]
predicted_nuclei_count = int(
image_patch_file_name_parser.
parse_image_patch_for_nuclei_count_prediction(
image_patch_file_name))
if predicted_nuclei_count >= highest_nuclei_count:
image_patch_path_with_highest_nuclei_count = image_patch_path
highest_nuclei_count = predicted_nuclei_count
image_patch_file = image_patch_path_with_highest_nuclei_count.split(
'/')[-1]
new_image_patch_path = output_diretory_path + "/" + image_patch_file
shutil.move(image_patch_path_with_highest_nuclei_count,
new_image_patch_path)
svs_loader = svs_l.SVSLoader(svs_input_folder_path)
path_utils.halt_script_if_path_does_not_exist(input_folder_path)
path_utils.create_directory_if_directory_does_not_exist_at_path(
output_folder_path)
most_salient_image_patch_case_directory_paths = path_utils.create_full_paths_to_directories_in_directory_path(
input_folder_path + "/visualizations/most_salient_image_patch_high_res")
CID_indexed_high_saliency_image_patch_metadata_objects_dict = image_patch_metadata_object_utils.case_directory_paths_containing_image_patches_to_dict_indexed_by_CID(
most_salient_image_patch_case_directory_paths)
high_nuclei_count_case_directory_paths = path_utils.create_full_paths_to_directories_in_directory_path(
input_folder_path_for_high_res_image_patches)
for high_nuclei_count_case_directory_path in high_nuclei_count_case_directory_paths:
high_nuclei_count_case_image_patches_paths = path_utils.create_full_paths_to_files_in_directory_path(
high_nuclei_count_case_directory_path)
first_image_patch_path = high_nuclei_count_case_image_patches_paths[0]
image_name = first_image_patch_path.split('/')[-1]
image_patch_metadata_object_with_high_nucleus = image_patch_file_name_parser.parse_image_patch_file_name_into_image_patch_metadata_object(
image_name)
corresponding_image_patch_metadata_object_with_highest_saliency = CID_indexed_high_saliency_image_patch_metadata_objects_dict[
image_patch_metadata_object_with_high_nucleus.case_id]
corresponding_image_patch_metadata_object_with_highest_saliency.image_patch_path = path_utils.create_full_paths_to_files_in_directory_path(
input_folder_path +
"/visualizations/most_salient_image_patch_high_res/" +
corresponding_image_patch_metadata_object_with_highest_saliency.case_id
)[0]
annotate_the_region_on_image_patch_with_the_highest_nuclei_count(
corresponding_image_patch_metadata_object_with_highest_saliency,
image_patch_metadata_object_with_high_nucleus, svs_loader,
svs_input_folder_path = args.svs_input_folder_path
csv_input_folder_path = args.csv_input_folder_path
output_folder_path = args.output_folder_path
accuracy_percentage_threshold = args.accuracy_percentage_threshold
path_utils.halt_script_if_path_does_not_exist(svs_input_folder_path)
path_utils.halt_script_if_path_does_not_exist(csv_input_folder_path)
path_utils.create_directory_if_directory_does_not_exist_at_path(
output_folder_path)
full_tcga_download_directory_paths = path_utils.create_full_paths_to_directories_in_directory_path(
svs_input_folder_path)
full_image_patch_data_dict_paths = path_utils.create_full_paths_to_files_in_directory_path(
csv_input_folder_path)
for full_tcga_download_directories_path_index, full_tcga_download_directory_path in enumerate(
full_tcga_download_directory_paths):
full_image_name_paths = path_utils.create_full_paths_to_files_in_directory_path(
full_tcga_download_directory_path)
image_name = full_image_name_paths[0].split('/')[-1][:-4]
output_path = output_folder_path + '/' + image_name + '/'
path_utils.create_directory_if_directory_does_not_exist_at_path(
output_path)
thumbnail = create_jpeg_thumbnail_of_wsi(full_image_name_paths[0])
thumbnail = thumbnail.convert("RGB")
thumbnail.save(output_path + image_name + "_original.jpeg", 'JPEG')
thumbnail = thumbnail.convert("RGBA")
svs_image = svs_utils.get_svs_image_of_wsi_from_path(
full_image_name_paths[0])
"the script will create it.",
required=True)
args = parser.parse_args()
input_folder_path = args.input_folder_path
output_folder_path = args.output_folder_path
path_utils.halt_script_if_path_does_not_exist(input_folder_path)
path_utils.create_directory_if_directory_does_not_exist_at_path(
output_folder_path)
case_directory_paths = path_utils.create_full_paths_to_directories_in_directory_path(
input_folder_path)
for case_directory_path in case_directory_paths:
full_image_patch_paths = path_utils.create_full_paths_to_files_in_directory_path(
case_directory_path)
for full_input_image_patch_path in full_image_patch_paths:
case_id = full_input_image_patch_path.split('/')[-2]
case_id_directory_path = output_folder_path + "/" + case_id
path_utils.create_directory_if_directory_does_not_exist_at_path(
case_id_directory_path)
image_patch_file = full_input_image_patch_path.split('/')[-1][:-4]
image_patch_root_path = case_id_directory_path + "/" + image_patch_file
path_utils.create_directory_if_directory_does_not_exist_at_path(
case_id_directory_path)
case_images_path = image_patch_root_path + "/" + "images"
path_utils.create_directory_if_directory_does_not_exist_at_path(
case_images_path)
parser.add_argument("-op", "--overlap_percentage", type=int, default=0,
help="Overlapping percentage between patches."
" Default value is 0.")
parser.add_argument("-ws ", "--window_size", type=int, default=10000,
help="Size for square window"
" Default value is 10000.")
args = parser.parse_args()
input_folder_path = args.input_folder_path
output_folder_path = args.output_folder_path
start_at_image_name = args.start_at_image_name
to_resolution_level = args.resolution_level
overlapping_percentage = float("{0:.2f}".format(args.overlap_percentage / 100))
window_size = args.window_size
path_utils.halt_script_if_path_does_not_exist(input_folder_path)
path_utils.create_directory_if_directory_does_not_exist_at_path(output_folder_path)
full_tcga_download_directories_paths = path_utils.create_full_paths_to_directories_in_directory_path(input_folder_path)
for full_tcga_download_directories_path in full_tcga_download_directories_paths:
full_image_name_paths = path_utils.create_full_paths_to_files_in_directory_path(full_tcga_download_directories_path)
for full_image_name_path in full_image_name_paths:
file_type = full_image_name_path[-4:]
if file_type == ".svs":
output_path = output_folder_path + '/'
svs_image_patch_extractor.split_to_jpeg_image_patches(full_image_name_path, output_path, to_resolution_level,
overlapping_percentage, window_size)
output_folder_path = args.output_folder_path
to_resolution_level = int(args.resolution_level)
overlapping_percentage = float("{0:.2f}".format(args.overlap_percentage / 100))
window_size = args.window_size
path_utils.halt_script_if_path_does_not_exist(svs_input_folder_path)
path_utils.halt_script_if_path_does_not_exist(csv_input_folder_path)
path_utils.create_directory_if_directory_does_not_exist_at_path(
output_folder_path)
tcga_download_directory_paths = path_utils.create_full_paths_to_directories_in_directory_path(
svs_input_folder_path)
case_image_patch_metadata_objects_csv_paths = path_utils.create_full_paths_to_files_in_directory_path(
csv_input_folder_path)
CID_indexed_image_patch_metadata_objects_dict = image_patch_metadata_object_utils.case_image_patch_metadata_csv_paths_to_dict_indexed_by_CID(
case_image_patch_metadata_objects_csv_paths)
for tcga_download_directories_path_index, tcga_download_directory_path in enumerate(
tcga_download_directory_paths):
full_image_name_paths = path_utils.create_full_paths_to_files_in_directory_path(
tcga_download_directory_path)
# there might be more than one image in a tcga download directory path (TO-DO: improve current solution)
first_image_name_path = full_image_name_paths[0]
case_id = first_image_name_path.split('/')[-1][:-4]
image_patch_metadata_objects_corresponding_to_CID = CID_indexed_image_patch_metadata_objects_dict[
case_id]
image_patch_metadata_object_with_highest_saliency_prediction = image_patch_metadata_object_utils.get_image_patch_metadata_object_with_the_highest_saliency(
input_folder_path = args.input_folder_path
output_folder_path = args.output_folder_path
labeled_dataset_output_folder_path = args.output_folder_path + "/labeled_SPOP_mutation_state_dataset"
mutation_file_directory_path = args.mutation_file_path
gene_of_interest = args.gene_of_interest
disc_operation = args.disc_operation
image_patch_input_folder_path = input_folder_path + "/image_patches_with_highest_nuclei_count/images_original"
path_utils.halt_script_if_path_does_not_exist(image_patch_input_folder_path)
path_utils.create_directory_if_directory_does_not_exist_at_path(
output_folder_path)
case_directory_paths = path_utils.create_full_paths_to_directories_in_directory_path(
image_patch_input_folder_path)
mutation_file_path = path_utils.create_full_paths_to_files_in_directory_path(
mutation_file_directory_path)[0]
all_image_patches_with_meta_info_paths = []
for case_directory_path in case_directory_paths:
print(case_directory_path)
image_patch_with_meta_info_path = path_utils.create_full_paths_to_files_in_directory_path(
case_directory_path)[0]
all_image_patches_with_meta_info_paths.append(
image_patch_with_meta_info_path)
gene_mutation_image_patch_labeler.label_image_patches(
all_image_patches_with_meta_info_paths, mutation_file_path,
gene_of_interest, labeled_dataset_output_folder_path, disc_operation)
copy_tree(input_folder_path + "/visualizations",
output_folder_path + "/visualizations")
images_input_folder_path = args.input_folder_path + '/' + "preprocessed_high_res_image_patches" + "/"
output_folder_path = args.output_folder_path
reorganized_image_patches = args.output_folder_path + '/' + "preprocessed_image_patches_organized_for_classifier" + "/"
image_mode = args.image_mode
path_utils.halt_script_if_path_does_not_exist(images_input_folder_path)
path_utils.create_directory_if_directory_does_not_exist_at_path(
output_folder_path)
case_directory_paths = path_utils.create_full_paths_to_directories_in_directory_path(
images_input_folder_path)
for case_directory_path in case_directory_paths:
full_image_patch_paths = None
if (image_mode == "rgb"):
full_image_patch_paths = path_utils.create_full_paths_to_files_in_directory_path(
case_directory_path + "/rgb")
elif (image_mode == "grayscale"):
full_image_patch_paths = path_utils.create_full_paths_to_files_in_directory_path(
case_directory_path + "/grayscale")
for full_input_image_patch_path in full_image_patch_paths:
case_id = full_input_image_patch_path.split('/')[-3]
case_id_directory_path = reorganized_image_patches + "/" + case_id
path_utils.create_directory_if_directory_does_not_exist_at_path(
case_id_directory_path)
image_patch_file = full_input_image_patch_path.split('/')[-1][:-4]
image_patch_root_path = case_id_directory_path + "/" + image_patch_file
path_utils.create_directory_if_directory_does_not_exist_at_path(
case_id_directory_path)
