Initial scripts for applying and evaluating baselines

Author: schilling40

scripts/baselines/cellpose-sam_IHC.py

@@ -0,0 +1,64 @@
+import json
+import os
+import time
+
+from cellpose import models, core, io
+from pathlib import Path
+from tqdm import trange
+from natsort import natsorted
+
+io.logger_setup()  # run this to get printing of progress
+
+# Check if colab notebook instance has GPU access
+if core.use_gpu() is False:
+    raise ImportError("No GPU access, change your runtime")
+
+model = models.CellposeModel(gpu=True)
+
+# *** change to your google drive folder path ***
+cochlea_dir = "/mnt/vast-nhr/projects/nim00007/data/moser/cochlea-lightsheet"
+input_dir = os.path.join(cochlea_dir, "AnnotatedImageCrops/F1ValidationIHCs")
+out_dir = os.path.join(cochlea_dir, "predictions/val_ihc/cellpose-sam")
+
+input_dir = Path(input_dir)
+if not input_dir.exists():
+    raise FileNotFoundError("directory does not exist")
+
+# *** change to your image extension ***
+image_ext = ".tif"
+
+# list all files
+files = natsorted([f for f in input_dir.glob("*"+image_ext) if "_masks" not in f.name and "_flows" not in f.name])
+
+if len(files) == 0:
+    raise FileNotFoundError("no image files found, did you specify the correct folder and extension?")
+else:
+    print(f"{len(files)} images in folder:")
+
+for f in files:
+    print(f.name)
+
+flow_threshold = 0.4
+cellprob_threshold = 0.0
+tile_norm_blocksize = 0
+
+masks_ext = ".png" if image_ext == ".png" else ".tif"
+for i in trange(len(files)):
+    f = files[i]
+    start = time.perf_counter()
+
+    img = io.imread(f)
+
+    basename = "".join(f.name.split(".")[:-1])
+    out_path = os.path.join(out_dir, f"{basename}_seg.tif")
+    timer_output = os.path.join(out_dir, f"{basename}_timer.json")
+
+    masks, flows, styles = model.eval(img, batch_size=32, flow_threshold=flow_threshold,
+                                      cellprob_threshold=cellprob_threshold,
+                                      normalize={"tile_norm_blocksize": tile_norm_blocksize})
+    io.imsave(out_path, masks)
+
+    duration = time.perf_counter() - start
+    time_dict = {"total_duration[s]": duration}
+    with open(timer_output, "w") as f:
+        json.dump(time_dict, f, indent='\t', separators=(',', ': '))

scripts/baselines/cellpose-sam_SGN.py

@@ -0,0 +1,64 @@
+import json
+import os
+import time
+
+from cellpose import models, core, io
+from pathlib import Path
+from tqdm import trange
+from natsort import natsorted
+
+io.logger_setup()  # run this to get printing of progress
+
+# Check if colab notebook instance has GPU access
+if core.use_gpu() is False:
+    raise ImportError("No GPU access, change your runtime")
+
+model = models.CellposeModel(gpu=True)
+
+# *** change to your google drive folder path ***
+cochlea_dir = "/mnt/vast-nhr/projects/nim00007/data/moser/cochlea-lightsheet"
+input_dir = os.path.join(cochlea_dir, "AnnotatedImageCrops/F1ValidationSGNs/for_consensus_annotation")
+out_dir = os.path.join(cochlea_dir, "predictions/val_sgn/cellpose-sam")
+
+input_dir = Path(input_dir)
+if not input_dir.exists():
+    raise FileNotFoundError("directory does not exist")
+
+# *** change to your image extension ***
+image_ext = ".tif"
+
+# list all files
+files = natsorted([f for f in input_dir.glob("*"+image_ext) if "_masks" not in f.name and "_flows" not in f.name])
+
+if len(files) == 0:
+    raise FileNotFoundError("no image files found, did you specify the correct folder and extension?")
+else:
+    print(f"{len(files)} images in folder:")
+
+for f in files:
+    print(f.name)
+
+flow_threshold = 0.4
+cellprob_threshold = 0.0
+tile_norm_blocksize = 0
+
+masks_ext = ".png" if image_ext == ".png" else ".tif"
+for i in trange(len(files)):
+    f = files[i]
+    start = time.perf_counter()
+
+    img = io.imread(f)
+
+    basename = "".join(f.name.split(".")[:-1])
+    out_path = os.path.join(out_dir, f"{basename}_seg.tif")
+    timer_output = os.path.join(out_dir, f"{basename}_timer.json")
+
+    masks, flows, styles = model.eval(img, batch_size=32, flow_threshold=flow_threshold,
+                                      cellprob_threshold=cellprob_threshold,
+                                      normalize={"tile_norm_blocksize": tile_norm_blocksize})
+    io.imsave(out_path, masks)
+
+    duration = time.perf_counter() - start
+    time_dict = {"total_duration[s]": duration}
+    with open(timer_output, "w") as f:
+        json.dump(time_dict, f, indent='\t', separators=(',', ': '))

scripts/baselines/cellpose3_IHC.py

@@ -0,0 +1,76 @@
+import json
+import os
+import time
+
+from cellpose import core, denoise, io, models
+from pathlib import Path
+from tqdm import trange
+from natsort import natsorted
+
+io.logger_setup()  # run this to get printing of progress
+
+# Check if colab notebook instance has GPU access
+if core.use_gpu() is False:
+    raise ImportError("No GPU access, change your runtime")
+
+model = models.CellposeModel(gpu=True)
+
+# *** change to your google drive folder path ***
+cochlea_dir = "/mnt/vast-nhr/projects/nim00007/data/moser/cochlea-lightsheet"
+input_dir = os.path.join(cochlea_dir, "AnnotatedImageCrops/F1ValidationIHCs")
+out_dir = os.path.join(cochlea_dir, "predictions/val_ihc/cellpose3")
+
+input_dir = Path(input_dir)
+if not input_dir.exists():
+    raise FileNotFoundError("directory does not exist")
+
+# *** change to your image extension ***
+image_ext = ".tif"
+
+# list all files
+files = natsorted([f for f in input_dir.glob("*"+image_ext) if "_masks" not in f.name and "_flows" not in f.name])
+
+if len(files) == 0:
+    raise FileNotFoundError("no image files found, did you specify the correct folder and extension?")
+else:
+    print(f"{len(files)} images in folder:")
+
+for f in files:
+    print(f.name)
+
+flow_threshold = 0.4
+cellprob_threshold = 0.0
+tile_norm_blocksize = 0
+
+masks_ext = ".png" if image_ext == ".png" else ".tif"
+for i in trange(len(files)):
+    f = files[i]
+    start = time.perf_counter()
+
+    img = io.imread(f)
+
+    basename = "".join(f.name.split(".")[:-1])
+    out_path = os.path.join(out_dir, f"{basename}_seg.tif")
+    timer_output = os.path.join(out_dir, f"{basename}_timer.json")
+
+    io.logger_setup()  # run this to get printing of progress
+
+    # DEFINE CELLPOSE MODEL
+    # model_type="cyto3" or "nuclei", or other model
+    # restore_type: "denoise_cyto3", "deblur_cyto3", "upsample_cyto3", "denoise_nuclei", "deblur_nuclei"
+    model = denoise.CellposeDenoiseModel(gpu=True, model_type="cyto3", restore_type="denoise_cyto3")
+
+    diameter = 20
+
+    masks, flows, styles, imgs_dn = model.eval(img, diameter=diameter, channels=[0, 0])
+
+    # masks, flows, styles = model.eval(img, batch_size=32, flow_threshold=flow_threshold,
+    #                                   cellprob_threshold=cellprob_threshold,
+    #                                   normalize={"tile_norm_blocksize": tile_norm_blocksize})
+
+    io.imsave(out_path, masks)
+
+    duration = time.perf_counter() - start
+    time_dict = {"total_duration[s]": duration}
+    with open(timer_output, "w") as f:
+        json.dump(time_dict, f, indent='\t', separators=(',', ': '))

scripts/baselines/cellpose3_SGN.py

@@ -0,0 +1,76 @@
+import json
+import os
+import time
+
+from cellpose import core, denoise, io, models
+from pathlib import Path
+from tqdm import trange
+from natsort import natsorted
+
+io.logger_setup()  # run this to get printing of progress
+
+# Check if colab notebook instance has GPU access
+if core.use_gpu() is False:
+    raise ImportError("No GPU access, change your runtime")
+
+model = models.CellposeModel(gpu=True)
+
+# *** change to your google drive folder path ***
+cochlea_dir = "/mnt/vast-nhr/projects/nim00007/data/moser/cochlea-lightsheet"
+input_dir = os.path.join(cochlea_dir, "AnnotatedImageCrops/F1ValidationSGNs/for_consensus_annotation")
+out_dir = os.path.join(cochlea_dir, "predictions/val_sgn/cellpose3")
+
+input_dir = Path(input_dir)
+if not input_dir.exists():
+    raise FileNotFoundError("directory does not exist")
+
+# *** change to your image extension ***
+image_ext = ".tif"
+
+# list all files
+files = natsorted([f for f in input_dir.glob("*"+image_ext) if "_masks" not in f.name and "_flows" not in f.name])
+
+if len(files) == 0:
+    raise FileNotFoundError("no image files found, did you specify the correct folder and extension?")
+else:
+    print(f"{len(files)} images in folder:")
+
+for f in files:
+    print(f.name)
+
+flow_threshold = 0.4
+cellprob_threshold = 0.0
+tile_norm_blocksize = 0
+
+masks_ext = ".png" if image_ext == ".png" else ".tif"
+for i in trange(len(files)):
+    f = files[i]
+    start = time.perf_counter()
+
+    img = io.imread(f)
+
+    basename = "".join(f.name.split(".")[:-1])
+    out_path = os.path.join(out_dir, f"{basename}_seg.tif")
+    timer_output = os.path.join(out_dir, f"{basename}_timer.json")
+
+    io.logger_setup()  # run this to get printing of progress
+
+    # DEFINE CELLPOSE MODEL
+    # model_type="cyto3" or "nuclei", or other model
+    # restore_type: "denoise_cyto3", "deblur_cyto3", "upsample_cyto3", "denoise_nuclei", "deblur_nuclei"
+    model = denoise.CellposeDenoiseModel(gpu=True, model_type="cyto3", restore_type="denoise_cyto3")
+
+    diameter = 20
+
+    masks, flows, styles, imgs_dn = model.eval(img, diameter=diameter, channels=[0, 0])
+
+    # masks, flows, styles = model.eval(img, batch_size=32, flow_threshold=flow_threshold,
+    #                                   cellprob_threshold=cellprob_threshold,
+    #                                   normalize={"tile_norm_blocksize": tile_norm_blocksize})
+
+    io.imsave(out_path, masks)
+
+    duration = time.perf_counter() - start
+    time_dict = {"total_duration[s]": duration}
+    with open(timer_output, "w") as f:
+        json.dump(time_dict, f, indent='\t', separators=(',', ': '))

scripts/baselines/distance-unet_IHC.py

@@ -0,0 +1,45 @@
+import os
+import sys
+
+script_dir = "/user/schilling40/u15000/flamingo-tools/scripts/prediction"
+sys.path.append(script_dir)
+
+import run_prediction_distance_unet
+
+checkpoint_dir = "/mnt/vast-nhr/projects/nim00007/data/moser/cochlea-lightsheet/trained_models/IHC"
+model_name = "v3_cochlea_distance_unet_IHC_supervised_2025-06-28"
+model_dir = os.path.join(checkpoint_dir, model_name)
+checkpoint = os.path.join(checkpoint_dir, model_name, "best.pt")
+
+cochlea_dir = "/mnt/vast-nhr/projects/nim00007/data/moser/cochlea-lightsheet"
+
+image_dir = os.path.join(cochlea_dir, "AnnotatedImageCrops/F1ValidationIHCs")
+
+out_dir = os.path.join(cochlea_dir, "predictions", "val_ihc", "distance_unet_v3")
+
+boundary_distance_threshold = 0.5
+seg_class = "ihc"
+
+block_shape = (128, 128, 128)
+halo = (16, 32, 32)
+
+block_shape_str = ",".join([str(b) for b in block_shape])
+halo_str = ",".join([str(h) for h in halo])
+
+images = [entry.path for entry in os.scandir(image_dir) if entry.is_file() and ".tif" in entry.path]
+
+for image in images:
+    sys.argv = [
+        os.path.join(script_dir, "run_prediction_distance_unet.py"),
+        f"--input={image}",
+        f"--output_folder={out_dir}",
+        f"--model={model_dir}",
+        f"--block_shape=[{block_shape_str}]",
+        f"--halo=[{halo_str}]",
+        "--memory",
+        "--time",
+        f"--seg_class={seg_class}",
+        f"--boundary_distance_threshold={boundary_distance_threshold}"
+    ]
+
+    run_prediction_distance_unet.main()

scripts/baselines/distance-unet_SGN.py

@@ -0,0 +1,41 @@
+import os
+import sys
+
+script_dir = "/user/schilling40/u15000/flamingo-tools/scripts/prediction"
+sys.path.append(script_dir)
+
+import run_prediction_distance_unet
+
+work_dir = "/mnt/lustre-grete/usr/u15000"
+checkpoint_dir = os.path.join(work_dir, "checkpoints")
+model_name = "cochlea_distance_unet_SGN_supervised_2025-05-27"
+model_dir = os.path.join(checkpoint_dir, model_name)
+checkpoint = os.path.join(checkpoint_dir, model_name, "best.pt")
+
+cochlea_dir = "/mnt/vast-nhr/projects/nim00007/data/moser/cochlea-lightsheet"
+
+image_dir = os.path.join(cochlea_dir, "AnnotatedImageCrops/F1ValidationSGNs/for_consensus_annotation")
+out_dir = os.path.join(cochlea_dir, "predictions/val_sgn")  # /distance_unet
+
+boundary_distance_threshold = 0.5
+seg_class = "sgn"
+
+block_shape = (128, 128, 128)
+halo = (16, 32, 32)
+
+images = [entry.path for entry in os.scandir(image_dir) if entry.is_file()]
+
+for image in images[:1]:
+    sys.argv = [
+        os.path.join(script_dir, "run_prediction_distance_unet.py"),
+        f"--input={image}",
+        f"--output_folder={out_dir}",
+        f"--model={model_dir}",
+        "--block_shape=[128,128,128]",
+        "--halo=[16,32,32]",
+        "--memory",
+        "--time",
+        f"--seg_class={seg_class}"
+    ]
+
+    run_prediction_distance_unet.main()