New Publication on Multi-Site Machine Learning for Brain Cancer Imaging

Read our latest paper in Nature Communications published in December 2022 regarding a large scale consortium effort to develop a federated learning approach for AI models based on brain cancer imaging. The segmented labels from our REMBRANDT MRI dataset are part of a world-wide federated platform Federated Tumor Segmentation (FeTS) that allows training specific machine learning models by leveraging information gathered from brain cancer datasets residing in collaborating sites without ever exchanging the data. Such a worldwide platform enables very large multi-site machine learning models in an effort to accelerate discovery.

In this paper, we describe how this Federated Learning model enabled big data for Rare Cancer Boundary Detection. This work was published with our collaborators Sypiron Bakas, Sarthak Pati, Ujjwal Baid et al at UPenn. Read a short summary of the paper in Inside Precision Medicine.