CSV file creation (links.csv) that maps file paths for the BraTS dataset.

• root_list: names of the directories containing the individual patient data.
  tot_list: lists of file paths for each patient (flair, seg, t1, t1ce, t2).

• Patient Directory Iterations:

  • os.listdir(root_df): to provide a list of patient directories.
  • os.path.join(root_df, filename_root): to create a sub-path for each patient directory.
  • np.sort(...): to sort the list of patient directories alphabetically.
  • tdqm(...): to provide a progress bar for the iteration.

• File Iterations for each Patient:

  • subpath = os.path.join(root_df, filename_root): to build the full path to the directory for a specific patient.
  • After which, the file paths for each patient are appended to the file_list list.
  • A for loop is used to iterate over the files in the patient directory.
  • os.path.join(subpath, filename): to create the full path to the file_list.

• Dataframe Creation:

  • pd.DataFrame(root_list, columns=['DIR']): to create a dataframe with the patient directory names.
  • pd.DataFrame(tot_list, columns=['flair', 'seg', 't1', 't1ce', 't2']): to create a dataframe with the file paths for each patient.
  • pd.concat(...): to concatenate the two dataframes into a single dataframe.
  • axis=1: to concatenate the dataframes along the columns.

root_list = [] tot_list = [] for filename_root in tqdm(np.sort(os.listdir(root_df))[:-2]): subpath = os.path.join(root_df, filename_root) file_list = [] for filename in np.sort(os.listdir(subpath)): file_list.append(os.path.join(subpath, filename)) root_list.append(filename_root) tot_list.append(file_list) maps = pd.concat([ pd.DataFrame(root_list, columns=['DIR']), pd.DataFrame(tot_list, columns=['flair', 'seg', 't1', 't1ce', 't2']) ], axis=1) maps.to_csv('scratch/links.csv', index=False)

Dictionary Population with file paths for different MRI modalities (seg, t1, t1ce, t2, flair) for each patient in the training dataset.

• image_path_dict: dictionary to store the file paths for each patient and their MRI modalities seg, flair, t1, t1ce, t2.