Upload m3d_cap_data_prepare.py

m3d_cap_data_prepare.py

@@ -0,0 +1,122 @@
+import os
+import numpy as np
+from PIL import Image
+import concurrent.futures
+from tqdm import tqdm
+from collections import Counter
+import unicodedata
+import monai.transforms as mtf
+from multiprocessing import Pool
+from unidecode import unidecode
+
+# input_dir = 'PATH/M3D_Cap/ct_quizze/'
+# output_dir = 'PATH/M3D_Cap_npy/ct_quizze/'
+
+input_dir = 'PATH/M3D_Cap/ct_case/'
+output_dir = 'PATH/M3D_Cap_npy/ct_case/'
+
+# Get all subfolders [00001, 00002....]
+subfolders = [folder for folder in os.listdir(input_dir) if os.path.isdir(os.path.join(input_dir, folder))]
+
+
+transform = mtf.Compose([
+    mtf.CropForeground(),
+    mtf.Resize(spatial_size=[32, 256, 256], mode="bilinear")
+])
+
+
+def process_subfolder(subfolder):
+    output_id_folder = os.path.join(output_dir, subfolder)
+    input_id_folder = os.path.join(input_dir, subfolder)
+
+    os.makedirs(output_id_folder, exist_ok=True)
+
+    for subsubfolder in os.listdir(input_id_folder):
+        if subsubfolder.endswith('.txt'):
+            text_path = os.path.join(input_dir, subfolder, subsubfolder)
+            with open(text_path, 'r') as file:
+                text_content = file.read()
+
+            search_text = "study_findings:"
+            index = text_content.find(search_text)
+
+            if index != -1:
+                filtered_text = text_content[index + len(search_text):].replace("\n", " ").strip()
+            else:
+                print("Specified string not found")
+                filtered_text = text_content.replace("\n", " ").strip()
+
+
+            if len(filtered_text.replace("\n", "").replace(" ", "")) < 5:
+                search_text = "discussion:"
+                index = text_content.find(search_text)
+                if index != -1:
+                    filtered_text = text_content[index + len(search_text):].replace("\n", " ").strip()
+                else:
+                    print("Specified string not found")
+                    filtered_text = text_content.replace("\n", " ").strip()
+
+
+            if len(filtered_text.replace("\n", "").replace(" ", "")) < 5:
+                filtered_text = text_content.replace("\n", " ").strip()
+
+
+            new_text_path = os.path.join(output_dir, subfolder, subsubfolder)
+            with open(new_text_path, 'w') as new_file:
+                new_file.write(filtered_text)
+
+        subsubfolder_path = os.path.join(input_dir, subfolder, subsubfolder)
+
+        if os.path.isdir(subsubfolder_path):
+            subsubfolder = unidecode(subsubfolder) # "Pöschl" -> Poschl
+            output_path = os.path.join(output_dir, subfolder, f'{subsubfolder}.npy')
+
+            image_files = [file for file in os.listdir(subsubfolder_path) if
+                           file.endswith('.jpeg') or file.endswith('.png')]
+
+            if len(image_files) == 0:
+                continue
+
+            image_files.sort(key=lambda x: int(os.path.splitext(x)[0]))
+
+            images_3d = []
+            for image_file in image_files:
+                image_path = os.path.join(subsubfolder_path, image_file)
+                try:
+                    img = Image.open(image_path)
+                    img = img.convert("L")
+                    img_array = np.array(img)
+                    # normalization
+                    img_array = img_array.astype(np.float32) / 255.0
+                    images_3d.append(img_array[None])
+                except:
+                    print("This image is error: ", image_path)
+
+            images_3d_pure = []
+            try:
+                img_shapes = [img.shape for img in images_3d]
+                item_counts = Counter(img_shapes)
+                most_common_shape = item_counts.most_common(1)[0][0]
+                for img in images_3d:
+                    if img.shape == most_common_shape:
+                        images_3d_pure.append(img)
+                final_3d_image = np.vstack(images_3d_pure)
+
+                image = final_3d_image[np.newaxis, ...]
+
+                image = image - image.min()
+                image = image / np.clip(image.max(), a_min=1e-8, a_max=None)
+
+                img_trans = transform(image)
+
+                np.save(output_path, img_trans)
+            except:
+                print([img.shape for img in images_3d])
+                print("This folder is vstack error: ", output_path)
+
+
+
+with Pool(processes=32) as pool:
+    with tqdm(total=len(subfolders), desc="Processing") as pbar:
+        for _ in pool.imap_unordered(process_subfolder, subfolders):
+            pbar.update(1)