Add link to paper, GitHub repository, and dataset description

README.md

@@ -1,7 +1,12 @@
 ---
-license: mit
 language:
 - en
+license: mit
+size_categories:
+- 100K<n<1M
+pretty_name: DeepSSIM Datasets
+task_categories:
+- other
 tags:
 - memorization
 - generative-models
@@ -9,7 +14,62 @@ tags:
 - medical-imaging
 - brain-mri
 - chest-xray
-pretty_name: DeepSSIM Datasets
-size_categories:
-- 100K<n<1M
----
+---
+
+# DeepSSIM Datasets
+
+[**Paper**](https://huggingface.co/papers/2509.16582) | [**GitHub**](https://github.com/brAIn-science/DeepSSIM)
+
+DeepSSIM is a novel self-supervised metric for quantifying memorization in generative models, particularly in the context of medical imaging. This repository contains the datasets and pre-computed artifacts used to evaluate DeepSSIM against state-of-the-art metrics.
+
+The evaluation involves synthetic brain MRI data generated by a Latent Diffusion Model (LDM) trained on 2,195 MRI scans from the publicly available **IXI** and **CoRR** datasets. It also includes evaluation on chest X-ray data.
+
+## Dataset Contents
+
+This repository provides:
+- **Similarity Matrices**: Pre-computed matrices for metrics including DeepSSIM, Chen, DAR, and SemDeDup.
+- **Index Files**: Required files to access matrix elements.
+- **Datasets**: Data used for training and testing generative models in memorization-prone conditions.
+
+## Usage
+
+The files in this repository are intended to be used with the scripts provided in the [official GitHub repository](https://github.com/brAIn-science/DeepSSIM).
+
+### Computing Similarity Matrices
+
+To compute a similarity matrix from embeddings:
+
+```bash
+python scripts/compute_matrix.py \
+  --dataset_images_dir PATH \
+  --embeddings_dir PATH \
+  --matrices_dir PATH \
+  --indices_dir PATH \
+  --model_path PATH \
+  --metric_name {deepssim, chen, dar, semdedup} \
+  --use_gpu
+```
+
+### Evaluating Metrics
+
+To evaluate a metric (computing macro F1 score, precision, and recall):
+
+```bash
+python scripts/eval.py \
+  --synth_indices_path PATH \
+  --real_indices_path PATH \
+  --matrix_path PATH \
+  --testset_csv PATH \
+  --metric_name {deepssim, chen, dar, semdedup}
+```
+
+## Citation
+
+```bibtex
+@article{scardace2025novel,
+    title={A Novel Metric for Detecting Memorization in Generative Models for Brain MRI Synthesis},
+    author={Scardace, Antonio and Puglisi, Lemuel and Guarnera, Francesco and Battiato, Sebastiano and Rav{\`\i}, Daniele},
+    journal={arXiv preprint arXiv:2509.16582},
+    year={2025}
+}
+```