Localdownload

DLICV/__main__.py

@@ -12,26 +12,62 @@
 warnings.simplefilter(action="ignore", category=FutureWarning)
 warnings.simplefilter(action="ignore", category=UserWarning)
 
+# VERSION = pkg_resources.require("NiChart_DLMUSE")[0].version
+VERSION = 1.0
 
 def main() -> None:
+    prog="DLICV"
     parser = argparse.ArgumentParser(
-        description="Use this to run inference with nnU-Net. This function is used when "
-        "you want to manually specify a folder containing a trained nnU-Net "
-        "model. This is useful when the nnunet environment variables "
-        "(nnUNet_results) are not set."
+        prog=prog,
+        description="DLICV - Deep Learning Intra Cranial Volume.",
+        usage = """
+        DLICV v{VERSION}
+        ICV calculation for structural MRI data.
+
+        Required arguments:
+            [-i, --in_dir]   The filepath of the input directory
+            [-o, --out_dir]  The filepath of the output directory
+        Optional arguments:
+            [-device]        cpu|cuda|mps - Depending on your system configuration (default: cuda)
+            [-h, --help]    Show this help message and exit.
+            [-V, --version] Show program's version number and exit.
+        EXAMPLE USAGE:
+            DLICV  -i           /path/to/input     \
+                   -o           /path/to/output    \
+                   -device      cpu|cuda|mps
+
+        """.format(VERSION=VERSION),
     )
+
+    # Required Arguments
     parser.add_argument(
         "-i",
         type=str,
         required=True,
-        help="input folder or list or image (nii.gz). Remember to use the correct channel numberings for your files (_0000 etc). File endings must be the same as the training dataset!",
+        help="[REQUIRED] Input folder with T1 sMRI images (nii.gz).",
     )
     parser.add_argument(
         "-o",
         type=str,
         required=True,
-        help="Output folder. If it does not exist it will be created. Predicted segmentations will "
-        "have the same name as their source images.",
+        help="[REQUIRED] Output folder. If it does not exist it will be created. Predicted segmentations will have the same name as their source images.",
+    )
+
+    # Optional Arguments
+    parser.add_argument(
+        "-device",
+        type=str,
+        default="cuda",
+        required=False,
+        help="[Recommended] Use this to set the device the inference should run with. Available options are 'cuda' (GPU), "
+        "'cpu' (CPU) or 'mps' (Apple M-series chips supporting 3D CNN).",
+    )
+    parser.add_argument(
+        "-V",
+        "--version",
+        action="version",
+        version=prog + ": v{VERSION}.".format(VERSION=VERSION),
+        help="Show the version and exit",
     )
     parser.add_argument(
         "-d",
@@ -144,15 +180,6 @@ def main() -> None:
         "5 and use -part_id 0, 1, 2, 3 and 4. Simple, right? Note: You are yourself responsible "
         "to make these run on separate GPUs! Use CUDA_VISIBLE_DEVICES (google, yo!)",
     )
-    parser.add_argument(
-        "-device",
-        type=str,
-        default="cuda",
-        required=False,
-        help="Use this to set the device the inference should run with. Available options are 'cuda' (GPU), "
-        "'cpu' (CPU) or 'mps' (Apple M-series chips supporting 3D CNN)."
-        "Use CUDA_VISIBLE_DEVICES=X nnUNetv2_predict [...] instead!",
-    )
     parser.add_argument(
         "--disable_progress_bar",
         action="store_true",
@@ -161,10 +188,33 @@ def main() -> None:
         help="Set this flag to disable progress bar. Recommended for HPC environments (non interactive "
         "jobs)",
     )
+    parser.add_argument(
+        "--clear_cache",
+        action="store_true",
+        required=False,
+        default=False,
+        help="Set this flag to clear any cached models before running. This is recommended if a previous download failed.",
+    )
 
     args = parser.parse_args()
     args.f = [0]
 
+    if args.clear_cache:
+        shutil.rmtree(os.path.join(
+            Path(__file__).parent,
+            "nnunet_results"
+        ))
+        shutil.rmtree(os.path.join(
+            Path(__file__).parent,
+            ".cache"
+        ))
+        if not args.i or not args.o:
+            print("Cache cleared and missing either -i / -o. Exiting.")
+            sys.exit(0)
+
+    if not args.i or not args.o:
+        parser.error("The following arguments are required: -i, -o")
+
     # data conversion
     src_folder = args.i  # input folder
     if not os.path.exists(args.o):  # create output folder if it does not exist
@@ -189,6 +239,16 @@ def main() -> None:
         % (args.d, args.d),
     )
 
+    if args.clear_cache:
+        shutil.rmtree(os.path.join(
+            Path(__file__).parent,
+            "nnunet_results"
+        ))
+        shutil.rmtree(os.path.join(
+            Path(__file__).parent,
+            ".cache"
+        ))
+
     # Check if model exists. If not exist, download using HuggingFace
     if not os.path.exists(model_folder):
         # HF download model
@@ -280,7 +340,7 @@ def main() -> None:
     if os.path.exists(des_folder):
         shutil.rmtree(des_folder)
 
-    print("Inference Process Done!")
+    print("DLICV Process Done!")
 
 
 if __name__ == "__main__":

README.md

@@ -27,6 +27,13 @@ Feel free to use it under the package's [licence](LICENCE)
 DLICV -i "input_folder" -o "output_folder" -device cpu
 ```
 
+#### Troubleshooting model download failures
+Our model download process creates several deep directory structures. If you are on Windows and your model download process fails, it may be due to Windows file path limitations. 
+
+To enable long path support in Windows 10, version 1607, and later, the registry key `HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\FileSystem LongPathsEnabled (Type: REG_DWORD)` must exist and be set to 1.
+
+If this affects you, we recommend re-running DLICV with the `--clear_cache` flag set on the first run.
+
 
 ## Contact
 For more information, please contact [CBICA Software](mailto:[email protected]).