FedNorm: Modality-Based Normalization in Federated Learning for Multi-Modal Liver Segmentation

Abstract

Given the high incidence and effective treatment options for liver diseases, they are of great socioeconomic importance. One of the most common methods for analyzing CT and MRI images for diagnosis and follow-up treatment is liver segmentation. Recent advances in deep learning have demonstrated encouraging results for automatic liver segmentation. Despite this, their success depends primarily on the availability of an annotated database, which is often not available because of privacy concerns. Federated Learning has been recently proposed as a solution to alleviate these challenges by training a shared global model on distributed clients without access to their local databases. Nevertheless, Federated Learning does not perform well when it is trained on a high degree of heterogeneity of image data due to multi-modal imaging, such as CT and MRI, and multiple scanner types. To this end, we propose FedNorm and its extension FedNorm+, two Federated Learning algorithms that use a modality-based normalization technique. Specifically, FedNorm normalizes the features on a client-level, while FedNorm+ employs the modality information of single slices in the feature normalization. Our methods were validated using 428 patients from six publicly available databases and compared to state-of-the-art Federated Learning algorithms and baseline models in heterogeneous settings (multi-institutional, multi-modal data). The experimental results demonstrate that our methods show an overall acceptable performance, achieve Dice per patient scores up to 0.961, consistently outperform locally trained models, and are on par or slightly better than centralized models.

MiFL: Multi-Input Neural Networks in Federated Learning

“…Unfortunately, this contribution is not directly linked to the Bio-Medical field. Bernecker et al 67 tackle the liver segmentation problem by proposing a multi-input normalization technique, which tries to encode CT and MRIs into a common representation.…”

Regularized Weight Aggregation in Networked Federated Learning for Glioblastoma Segmentation

“…FL has been combined with domain adaptation [25], contrastive learning [9], and knowledge distillation [13] in order to learn a more generalizable federated model. Other limitations include cross domain and imbalance of annotated data (limited labeling budgets) [6]. The challenge of data heterogeneity and domain shifting was recently tackled in novel ways by, for example, federated disentanglement learning via disentangling the parameter space into shape and appearance [6] and automated federated averaging based on Dirichlet distribution [22].…”

“…Other limitations include cross domain and imbalance of annotated data (limited labeling budgets) [6]. The challenge of data heterogeneity and domain shifting was recently tackled in novel ways by, for example, federated disentanglement learning via disentangling the parameter space into shape and appearance [6] and automated federated averaging based on Dirichlet distribution [22]. Dynamic Re-Weighting mechanisms [12], federated cross ensemble learning [24], and label-agnostic (mixed labels) unified FL formed by a mixture of the client distributions [21] have been recently proposed to relax an unrealistic assumption that each client's training set will be annotated similarly and therefore follows the same image supervision level during the training of an image segmentation model.…”

Federated Learning for Medical Applications: A Taxonomy, Current Trends, Challenges, and Future Research Directions

“…FL-based solutions recently gained popularity by training the shared global models by distributed clients with heterogeneous image datasets. Bernecker et al [199] proposed two FL algorithms, namely FedNorm and FedNorm+, that were based on modality-based normalization techniques. They validated their method on the multi-modal and multi-institutional datasets (6 centers, 428 patients).…”