Update README.md

README.md

@@ -61,6 +61,14 @@ The model was trained on the **IDUN HPC Cluster**.
 - **Carbon Tracking:** Training included environmental footprint monitoring (~0.25 kW/GPU on a low-carbon Norwegian grid).
 - **Stability:** Evaluation shows consistent convergence across 5 folds, validated by a rigorous audit of the patient split.
 
+## Explainability (Grad-CAM)
+The model includes a wrapper for Explainable AI. It targets the layer4 of the T2 encoder to generate 3D heatmaps, 
+helping to identify the specific anatomical regions influencing the "Malignant" prediction.
+
+## Evaluation Results
+The primary metric used is ROC AUC (Area Under the Receiver Operating Analytic Curve).
+Detailed performance graphs and confusion matrices for the ensemble can be found in the associated [GitHub Repository](https://github.com/THOUAN-Simon/ODELIA_Challenge_CV_DL)
+
 ## How to Load the Model
 This model requires the `MultiStreamResNet` class from the `utils.py` file.
 
@@ -76,10 +84,6 @@ model.load_state_dict(torch.load("best_resnet_odelia_fold0.pth", map_location="c
 model.eval()
 ```
 
-## Explainability (Grad-CAM)
-The model includes a wrapper for Explainable AI. It targets the layer4 of the T2 encoder to generate 3D heatmaps, 
-helping to identify the specific anatomical regions influencing the "Malignant" prediction.
-
 ## Limitations & Ethical Considerations
 This model is for research purposes only. It was trained on the ODELIA proprietary dataset. 
 Predictions should not be used for clinical diagnosis without professional medical supervision.