|
|
import os |
|
|
import pydicom |
|
|
import nibabel as nib |
|
|
import pandas as pd |
|
|
import dicom2nifti |
|
|
from tqdm import tqdm |
|
|
import concurrent.futures |
|
|
|
|
|
def get_knee_side(dicom_path): |
|
|
""" |
|
|
Reads DICOM metadata and extracts the knee side from SeriesDescription. |
|
|
Returns: |
|
|
int: 1 for RIGHT, 2 for LEFT (based on SeriesDescription), or None on error. |
|
|
""" |
|
|
try: |
|
|
ds = pydicom.dcmread(dicom_path, stop_before_pixels=True) |
|
|
series_desc = getattr(ds, "SeriesDescription", "").upper() |
|
|
print(series_desc) |
|
|
return 'RIGHT' if series_desc == "SAG_3D_DESS_RIGHT" else 'LEFT' |
|
|
except Exception as e: |
|
|
print(f"Error reading DICOM file {dicom_path}: {e}") |
|
|
return None |
|
|
|
|
|
def convert_dcm_to_nifti(dicom_path, save_image_filepath): |
|
|
""" |
|
|
Converts DICOM files to NIfTI format. |
|
|
Args: |
|
|
dicom_path (str): Path to the DICOM file. |
|
|
save_image_filepath (str): Path to save the NIfTI file. |
|
|
""" |
|
|
dicom2nifti.dicom_series_to_nifti(dicom_path, save_image_filepath, reorient_nifti=True) |
|
|
print(f"Saved {save_image_filepath}") |
|
|
return save_image_filepath |
|
|
|
|
|
def process_row(row, save_image_folder): |
|
|
""" |
|
|
Process a single row from the DataFrame. |
|
|
""" |
|
|
patient_id = row['patient_id'] |
|
|
time = row['time'] |
|
|
study_folder = row['study_folder'] |
|
|
series_folder = row['series_folder'] |
|
|
knee_side = row['knee_side'] |
|
|
path = row['path'] |
|
|
|
|
|
save_image_filepath = os.path.join(save_image_folder, str(patient_id), str(time), |
|
|
f'{study_folder}_{series_folder}_{knee_side}.nii.gz') |
|
|
|
|
|
if os.path.exists(save_image_filepath): |
|
|
print(f"Skipping {save_image_filepath} because it already exists") |
|
|
else: |
|
|
os.makedirs(os.path.dirname(save_image_filepath), exist_ok=True) |
|
|
convert_dcm_to_nifti(path, save_image_filepath) |
|
|
|
|
|
row_copy = row.copy() |
|
|
row_copy['save_image_filepath'] = save_image_filepath |
|
|
return row_copy |
|
|
|
|
|
if __name__ == "__main__": |
|
|
save_image_folder = '/data/images' |
|
|
df = pd.read_csv('/data/all_studies.csv') |
|
|
|
|
|
|
|
|
max_workers = 8 |
|
|
save_rows = [] |
|
|
|
|
|
batch_size = 100 |
|
|
for i in range(0, len(df), batch_size): |
|
|
batch_df = df.iloc[i:i+batch_size] |
|
|
|
|
|
with concurrent.futures.ThreadPoolExecutor(max_workers=max_workers) as executor: |
|
|
futures = [executor.submit(process_row, row, save_image_folder) |
|
|
for _, row in batch_df.iterrows()] |
|
|
|
|
|
for future in tqdm(concurrent.futures.as_completed(futures), |
|
|
total=len(futures), |
|
|
desc=f"Batch {i//batch_size+1}/{(len(df)+batch_size-1)//batch_size}"): |
|
|
try: |
|
|
result = future.result() |
|
|
save_rows.append(result) |
|
|
except Exception as e: |
|
|
print(f"Processing error: {e}") |
|
|
|
