First commit

README.md

@@ -1,14 +1 @@
----
-title: FLAMeS
-emoji: 🏃
-colorFrom: blue
-colorTo: red
-sdk: gradio
-sdk_version: 5.9.1
-app_file: app.py
-pinned: false
-license: gpl-3.0
-short_description: Automated segmentation of MS lesions
----
-
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+Tool for the automated segmentation of multiple sclerosis lesions from diverse FLAIR scans.

app.py

@@ -0,0 +1,59 @@
+import gradio as gr
+import subprocess
+import os
+import shutil
+
+# Paths
+INPUT_DIR = "/tmp/input"
+OUTPUT_DIR = "/tmp/output"
+MODEL_DIR = "./model"
+
+def run_nnunet_predict(nifti_file):
+    # Prepare directories
+    os.makedirs(INPUT_DIR, exist_ok=True)
+    os.makedirs(OUTPUT_DIR, exist_ok=True)
+
+    # Save the uploaded file to the input directory
+    input_path = os.path.join(INPUT_DIR, "image.nii.gz")
+    shutil.copy(nifti_file.name, input_path)
+
+    # Set environment variables for nnUNet
+    os.environ["nnUNet_raw"] = MODEL_DIR
+    os.environ["nnUNet_preprocessed"] = MODEL_DIR
+    os.environ["nnUNet_results"] = MODEL_DIR
+
+    # Construct the nnUNetv2_predict command
+    command = [
+        "nnUNetv2_predict",
+        "-i", INPUT_DIR,
+        "-o", OUTPUT_DIR,
+        "-d", "004",                  # Dataset ID
+        "-c", "3d_fullres",           # Configuration
+        "-tr", "nnUNetTrainer_8000epochs",  # Trainer name
+    ]
+
+    # Run the command
+    try:
+        subprocess.run(command, check=True)
+        # Get the output file
+        output_file = os.path.join(OUTPUT_DIR, "image.nii.gz")
+        return output_file
+    except subprocess.CalledProcessError as e:
+        return f"Error: {e}"
+
+# Gradio Interface
+interface = gr.Interface(
+    fn=run_nnunet_predict,
+    inputs=gr.File(label="Upload FLAIR Image (.nii.gz)"),
+    outputs=gr.File(label="Download Segmentation Mask"),
+    title="FLAMeS: FLAIR Lesion Analysis in Multiple Sclerosis",
+    description=(
+        "Upload a skull-stripped FLAIR image (.nii.gz) to generate a binary segmentation of MS lesions. "
+        "This model uses nnUNetv2 for inference with ensemble predictions."
+    ),
+)
+
+# Launch the app
+if __name__ == "__main__":
+    interface.launch()
+

requirements.txt

@@ -0,0 +1,5 @@
+gradio
+torch
+nnunetv2
+nibabel
+numpy