Reproducibility of block extraction and SGN post-processing

flamingo_tools/extract_block.py

@@ -0,0 +1,98 @@
+import os
+from typing import Optional, List
+
+import imageio.v3 as imageio
+import numpy as np
+import zarr
+
+import flamingo_tools.s3_utils as s3_utils
+from flamingo_tools.file_utils import read_image_data
+
+
+def extract_block(
+    input_path: str,
+    coords: List[int],
+    output_dir: Optional[str] = None,
+    input_key: Optional[str] = None,
+    output_key: Optional[str] = None,
+    resolution: float = 0.38,
+    roi_halo: List[int] = [128, 128, 64],
+    tif: bool = False,
+    s3: Optional[bool] = False,
+    s3_credentials: Optional[str] = None,
+    s3_bucket_name: Optional[str] = None,
+    s3_service_endpoint: Optional[str] = None,
+):
+    """Extract block around coordinate from input data according to a given halo.
+    Either from a local file or from an S3 bucket.
+
+    Args:
+        input_path: Input folder in n5 / ome-zarr format.
+        coords: Center coordinates of extracted 3D volume.
+        output_dir: Output directory for saving output as <basename>_crop.n5. Default: input directory.
+        input_key: Input key for data in input file.
+        output_key: Output key for data in n5 output or used as suffix for tif output.
+        roi_halo: ROI halo of extracted 3D volume.
+        tif: Flag for tif output
+        s3: Flag for considering input_path for S3 bucket.
+        s3_bucket_name: S3 bucket name.
+        s3_service_endpoint: S3 service endpoint.
+        s3_credentials: File path to credentials for S3 bucket.
+    """
+    coords = [int(round(c)) for c in coords]
+    coord_string = "-".join([str(c).zfill(4) for c in coords])
+
+    # Dimensions are inversed to view in MoBIE (x y z) -> (z y x)
+    coords.reverse()
+    roi_halo.reverse()
+
+    input_content = list(filter(None, input_path.split("/")))
+
+    if s3:
+        image_name = input_content[-1].split(".")[0]
+        if len(image_name.split("_")) > 1:
+            resized_suffix = "_resized"
+            image_prefix = image_name.split("_")[0]
+        else:
+            resized_suffix = ""
+            image_prefix = image_name
+        basename = input_content[0] + resized_suffix
+    else:
+        basename = "".join(input_content[-1].split(".")[:-1])
+        image_prefix = basename.split("_")[-1]
+
+    input_dir = input_path.split(basename)[0]
+    input_dir = os.path.abspath(input_dir)
+
+    if output_dir == "":
+        output_dir = input_dir
+
+    if tif:
+        if output_key is None:
+            output_name = basename + "_crop_" + coord_string + "_" + image_prefix + ".tif"
+        else:
+            output_name = basename + "_" + image_prefix + "_crop_" + coord_string + "_" + output_key + ".tif"
+
+        output_file = os.path.join(output_dir, output_name)
+    else:
+        output_key = "raw" if output_key is None else output_key
+        output_file = os.path.join(output_dir, basename + "_crop_" + coord_string + ".n5")
+
+    coords = np.array(coords)
+    coords = coords / resolution
+    coords = np.round(coords).astype(np.int32)
+
+    roi = tuple(slice(co - rh, co + rh) for co, rh in zip(coords, roi_halo))
+
+    if s3:
+        input_path, fs = s3_utils.get_s3_path(input_path, bucket_name=s3_bucket_name,
+                                              service_endpoint=s3_service_endpoint, credential_file=s3_credentials)
+
+    data_ = read_image_data(input_path, input_key)
+    data_roi = data_[roi]
+
+    if tif:
+        imageio.imwrite(output_file, data_roi, compression="zlib")
+    else:
+        with zarr.open(output_file, mode="w") as f_out:
+            f_out.create_dataset(output_key, data=data_roi, compression="gzip")

reproducibility/README.md

@@ -0,0 +1,23 @@
+# Scripts to ensure/facilitate the reproducbility of pre-/post-processing
+
+The scripts within these folders document steps involved in the pre-/post-processing or analysis of data.
+The folders contain a script for reproducibility as well as dictionaries in JSON format, which contain script parameters and additional information.
+
+## Extraction of blocks from a 3D volume
+
+The extraction of blocks from a 3D volume is required for the creation of annotations, empty crops, or others regions of interest.
+
+Usage:
+```
+python repro_block_extraction.py --input <JSON-file> --output <out-dir>
+``` 
+
+## Post-processing of SGN segmentation
+
+The post-processing of the SGN segmentation may involve the erosion of the segmentation to exclude artifacts, the variation of the minimal number of nodes within a component, or the minimal distance between nodes to consider them the same component.
+
+Usage:
+ ```
+python repro_postprocess_sgn_v1.py --input <JSON-file> --output <out-dir>
+``` 
+

reproducibility/block_extraction/2025-05-SGN_annotated_crops.json

@@ -0,0 +1,147 @@
+[
+	{
+		"cochlea": "M_LR_000144_L",
+		"image_channel": "PV_resized",
+		"crop_centers": [
+			[
+				1040,
+				700,
+				1390
+			],
+			[
+				1300,
+				690,
+				1191
+			],
+			[
+				1455,
+				814,
+				1191
+			]
+		],
+		"halo_size": [
+			128,
+			128,
+			32
+		]
+	},
+	{
+		"cochlea": "M_LR_000145_L",
+		"image_channel": "PV_resized",
+		"crop_centers": [
+			[
+				972,
+				862,
+				610
+			],
+			[
+				1301,
+				812,
+				740
+			]
+		],
+		"halo_size": [
+			128,
+			128,
+			32
+		]
+	},
+	{
+		"cochlea": "M_LR_000167_R",
+		"image_channel": "PV",
+		"crop_centers": [
+			[
+				909,
+				813,
+				810
+			],
+			[
+				916,
+				800,
+				722
+			],
+			[
+				951,
+				742,
+				652
+			],
+			[
+				951,
+				742,
+				602
+			],
+			[
+				1251,
+				266,
+				575
+			],
+			[
+				1210,
+				148,
+				606
+			],
+			[
+				1305,
+				341,
+				526
+			],
+			[
+				1137,
+				669,
+				1044
+			]
+		],
+		"halo_size": [
+			128,
+			128,
+			32
+		]
+	},
+	{
+		"cochlea": "M_LR_000151_R",
+		"image_channel": "PV_resized",
+		"crop_centers": [
+			[
+				1275,
+				806,
+				578
+			],
+			[
+				875,
+				539,
+				747
+			],
+			[
+				1300,
+				535,
+				747
+			]
+		],
+		"halo_size": [
+			128,
+			128,
+			32
+		]
+	},
+	{
+		"cochlea": "M_LR_000184_L",
+		"image_channel": "PV_resized",
+		"crop_centers": [
+			[
+				1326,
+				844,
+				516
+			],
+			[
+				885,
+				913,
+				1023
+			]
+		],
+		"halo_size": [
+			128,
+			128,
+			32
+		]
+	}
+]

reproducibility/block_extraction/2025-05-SGN_domain_crops.json

@@ -0,0 +1,102 @@
+[
+	{
+		"cochlea": "M_LR_000144_L",
+		"image_channel": "PV_resized",
+		"crop_centers": [
+			[
+				1490,
+				600,
+				735
+			],
+			[
+				1540,
+				930,
+				865
+			]
+		],
+		"halo_size": [
+			256,
+			256,
+			128
+		]
+	},
+	{
+		"cochlea": "M_LR_000145_L",
+		"image_channel": "PV_resized",
+		"crop_centers": [
+			[
+				1300,
+				700,
+				700
+			]
+		],
+		"halo_size": [
+			256,
+			256,
+			128
+		]
+	},
+	{
+		"cochlea": "M_LR_000151_R",
+		"image_channel": "PV_resized",
+		"crop_centers": [
+			[
+				1350,
+				690,
+				1000
+			],
+			[
+				1350,
+				570,
+				734
+			]
+		],
+		"halo_size": [
+			256,
+			256,
+			128
+		]
+	},
+	{
+		"cochlea": "M_LR_000155_L",
+		"image_channel": "PV_resized",
+		"crop_centers": [
+			[
+				900,
+				1090,
+				700
+			],
+			[
+				1464,
+				830,
+				420
+			]
+		],
+		"halo_size": [
+			256,
+			256,
+			128
+		]
+	},
+	{
+		"cochlea": "M_LR_000167_R",
+		"image_channel": "PV",
+		"crop_centers": [
+			[
+				1360,
+				1040,
+				840
+			],
+			[
+				900,
+				780,
+				780
+			]
+		],
+		"halo_size": [
+			256,
+			256,
+			128
+		]
+	}
+]