Extract tiles from SVS

preprocessing/svs_to_random_tiles.py

+# %% 
+import gc
+import os
+import sys
+from pathlib import Path
+
+import numpy as np
+import pandas as pd
+
+# custom Histology library
+from histo_lib import wsi
+from inf.config import SVS_DIR
+
+# %%
+#PATH = Path('..')
+
+SIZE = 512
+SIZE_NO_OVERLAP = 1024
+
+LEVEL = 0
+
+N_TILES_MAX = 100
+
+OUT_DIR = SVS_DIR.parents[0] / f'tiles_level_{0}' # folder in the same parent folder of SVS_DIR
+os.makedirs(OUT_DIR, exist_ok=True)
+
+# %%
+svs_files = os.listdir(SVS_DIR)
+# %%
+wrong_svs = []
+for index, SVS in enumerate(svs_files): #for each svs file
+
+    #sys.stdout.flush()
+    sys.stdout.write(f'\r Processing svs: {index+1}/{len(svs_files)} : {SVS}')
+
+    try:
+        img = wsi.open_wsi(f'{SVS_DIR}/{SVS}') #open the image
+
+        prefix = f'{OUT_DIR}/{SVS}_' 
+        # prefix =  f'{OUTDIR}/{SVS}/{SVS}' if you want to create a subfolder for each wsi
+
+        # get tissue box
+        box_tissue = wsi.get_tissue_region(img) # get the tissue box
+        w = box_tissue[2] - box_tissue[0] #get height of the tissue box
+        h = box_tissue[3] - box_tissue[1] #get width of the tissue box
+
+        #compute the maximum number of tiles that can be extracted
+        min_size = min(h,w)
+        max_tiles = (min_size // SIZE_NO_OVERLAP)**2
+
+        # generate random tiles
+        n_tiles = min(N_TILES_MAX, max_tiles)
+        wsi.wsi2random_tiles(img, SIZE, LEVEL, n_tiles, check_tissue=True, prefix=prefix, max_iter=2*N_TILES_MAX)
+
+        #clean workspace
+        del img
+        gc.collect()
+        
+        
+    except Exception as e:
+        wrong_svs.append(SVS)
+        print(e)
+
+
+# %%