Update README.md

README.md

@@ -1,33 +1,87 @@
-# Robust Breast MRI Classification via Multi-Sequence Attention-Based Multiple Instance Learning and Focal Loss Optimization
-This repository contains code for **Attention-Based Multiple Instance Learning (ABMIL)** models for **breast MRI classification**.  
+---
+license_link: https://github.com/adarshbhandaryp/ABMIL
+datasets:
+- ODELIA-AI/ODELIA-Challenge-2025
+language:
+- en
+metrics:
+- roc_auc
+pipeline_tag: image-classification
+tags:
+- breast
+- cancer
+- mri
+- odelia
+- abmil
+- swin-transformer
+extra_gated_prompt: >-
+  ### Model Usage Agreement
+
+  By accessing or using this model (the "Model"), you acknowledge and agree
+  to the following terms:
 
-# 🩺 Code for ODELIA Breast MRI Challenge 2025
-## Overview  
-The **ODELIA Breast MRI Challenge 2025** benchmarks subject-level classification on a multi-center dataset of unilateral breast MRI in NIfTI format with three labels: **normal, benign, and malignant**. The goal is to advance clinically meaningful AI by tackling class imbalance, temporal contrast dynamics, and multi-sequence integration. Participants build models that ingest pre- and post-contrast sequences (and subtraction) and produce patient-level predictions evaluated on robust metrics. The challenge emphasizes reproducibility and fairness across institutions, encouraging architectures that learn from full 3D volumes without lesion annotations. Results aim to accelerate real-world deployment of reliable screening and diagnostic support systems across heterogeneous clinical settings.
+  1. The Model is provided for non-commercial academic and research use only.
+  2. The Model must not be used for diagnosis, treatment, clinical decision
+     making, or any application involving real-patient care.
+  3. Do not identify, re-identify, or deanonymize people whose data may have
+     contributed to the training or evaluation data.
+  4. Publications or derivative research must acknowledge the ODELIA
+     consortium and cite the ABMIL authors and source repository.
+  5. You are responsible for validating outputs in your research context.
+     The Model is provided as-is, without warranties.
 
-## Method  
-We present a multi-sequence **Attention-Based Multiple Instance Learning (ABMIL)** framework for subject-level breast MRI classification. Each breast volume is represented by 32 axial slices with **pre-contrast, early post-contrast, and on-the-fly subtraction** channels. Slice features are extracted with **Swin Transformer Tiny** (ImageNet-pretrained) and aggregated via **gated attention** to form a patient-level embedding. To mitigate pronounced class imbalance, we optimize with **focal loss** and found **γ = 5.0** consistently strongest across validation and pre-evaluation tests. The approach operates without lesion annotations, leverages temporal enhancement patterns, and delivers robust performance across institutions—tailored for scalable, clinically relevant evaluation.
+  This gated access does not grant rights beyond the license and permissions
+  stated in this model card. Third-party components retain their own licenses.
+---
 
-# Code
+# ODELIA ABMIL Model
 
+Attention-Based Multiple Instance Learning (ABMIL) model for subject-level
+three-class breast MRI classification in the ODELIA Challenge 2025.
 
-The training pipeline is controlled by **CSV configuration files** located in the [`config/`](config/) folder.
+This model was contributed by the ABMIL team. The source implementation is
+available at [adarshbhandaryp/ABMIL](https://github.com/adarshbhandaryp/ABMIL).
 
----
+## Method
+
+The model represents each subject using 32 axial slices with pre-contrast,
+early post-contrast, and subtraction channels. A Swin Transformer Tiny
+backbone extracts slice features. Gated attention aggregates the slice
+features into a patient-level representation, followed by a three-class
+classifier. The source README reports focal loss with gamma 5.0.
+
+## Intended use and limitations
+
+This release is for research and benchmarking only. It is not a medical
+device, is not clinically validated, and must not be used for diagnosis,
+prognosis, treatment, triage, or other clinical decisions. Performance may
+vary across institutions, scanners, acquisition protocols, and populations.
 
-## 🚀 Running the Code
+## Setup
+
+The source repository provides the complete training and evaluation code and
+requirements:
 
 ```bash
+git clone https://github.com/adarshbhandaryp/ABMIL.git
+cd ABMIL
 pip install -r requirements.txt
-
-python main.py \
-  --configuration_file "config/config_mil_cross_nii.csv" \
-  --project_name "Breast MRI MIL Cross Attention"
-configuration_file → Path to one of the CSV configs in config/.
-project_name → Custom name for logging runs in Weights & Biases
 ```
 
-## Contact Us  
-- **Dr. Tri-Thien Nguyen** – [tri-thien.nguyen@fau.de](mailto:tri-thien.nguyen@fau.de) · [Profile](https://lme.tf.fau.de/person/ttnguyen)  
-- **Adarsh Bhandary Panambur** – [adarsh.bhandary.panambur@fau.de](mailto:adarsh.bhandary.panambur@fau.de) · [Profile](https://lme.tf.fau.de/person/panambur/)  
-- **Pattern Recognition Lab (FAU Erlangen-Nürnberg)** – [https://lme.tf.fau.de/](https://lme.tf.fau.de/)
+Refer to the source repository for configuration-file usage and preprocessing
+requirements. The model expects ODELIA-style multi-sequence breast MRI input.
+
+## License and attribution
+
+This repository is published under the ODELIA and ABMIL authors' permission.
+
+The Swin Transformer implementation and pretrained ImageNet weights are
+third-party components. Their applicable licenses and terms must be preserved
+and reviewed before redistribution. See the dependency notices in the source
+repository and the relevant upstream package/model licenses.
+
+Please cite or acknowledge:
+
+- ODELIA Challenge 2025 and the ODELIA consortium.
+- The ABMIL authors and [source repository](https://github.com/adarshbhandaryp/ABMIL).
+- The ODELIA paper: [arXiv:2506.00474](https://arxiv.org/abs/2506.00474).