IPMN Risk Assessment under Federated Learning Paradigm
Hongyi Pan, Ziliang Hong, Gorkem Durak, Elif Keles, Halil Ertugrul Aktas, Yavuz Taktak, Alpay Medetalibeyoglu, Zheyuan Zhang, Yury Velichko, Concetto Spampinato, Ivo Schoots, Marco J. Bruno, Pallavi Tiwari, Candice Bolan, Tamas Gonda, Frank Miller, Rajesh N. Keswani, Michael B. Wallace, Ziyue Xu, Ulas Bagci
TL;DR
This work tackles IPMN risk classification from pancreas MRI under data privacy constraints by presenting a large multi-center dataset and a privacy-preserving federated learning framework. A 3D DenseNet-121 backbone is evaluated for both centralized and federated training across seven institutions, using T1- and T2-weighted MRI images cropped to ROIs. Federated approaches with FedAvg and FedProx demonstrate high classification performance, closely matching centralized baselines while preserving patient privacy, highlighting the feasibility of secure cross-institution collaboration. The study provides dataset benchmarks, methodological comparisons, and practical evidence that privacy-preserving federated learning can enhance IPMN classification in real-world clinical settings, potentially enabling scalable deployment across centers. ${F_k(oldsymbol{w})} = rac{1}{N_k} \sum_{i \in \mathcal{D}_k} \ell(f(\mathbf{x}_i; \boldsymbol{w}), \mathbf{y}_i)$ and $\boldsymbol{w} = \frac{\sum_k N_k \boldsymbol{w}_k}{\sum_k N_k}$, with ${F_k^{\text{prox}}(\boldsymbol{w})} = F_k(\boldsymbol{w}) + \frac{\mu}{2} \| \boldsymbol{w} - \boldsymbol{w}_t \|^2$ illustrating the core federated optimization mechanisms.
Abstract
Accurate classification of Intraductal Papillary Mucinous Neoplasms (IPMN) is essential for identifying high-risk cases that require timely intervention. In this study, we develop a federated learning framework for multi-center IPMN classification utilizing a comprehensive pancreas MRI dataset. This dataset includes 652 T1-weighted and 655 T2-weighted MRI images, accompanied by corresponding IPMN risk scores from 7 leading medical institutions, making it the largest and most diverse dataset for IPMN classification to date. We assess the performance of DenseNet-121 in both centralized and federated settings for training on distributed data. Our results demonstrate that the federated learning approach achieves high classification accuracy comparable to centralized learning while ensuring data privacy across institutions. This work marks a significant advancement in collaborative IPMN classification, facilitating secure and high-accuracy model training across multiple centers.