Multi-Site Class-Incremental Learning with Weighted Experts in Echocardiography

TL;DR

This paper introduces a multi-site class-incremental learning approach for echocardiography view classification that combines multiple expert networks with weighted score fusion, reducing training time and handling data privacy issues.

Contribution

It presents a novel method of learning expert networks for each dataset and combining them with weighted score fusion, addressing catastrophic forgetting and data sharing constraints.

Findings

Significant reduction in training time.

Improved view classification accuracy.

Effective handling of multi-site data variations.

Abstract

Building an echocardiography view classifier that maintains performance in real-life cases requires diverse multi-site data, and frequent updates with newly available data to mitigate model drift. Simply fine-tuning on new datasets results in "catastrophic forgetting", and cannot adapt to variations of view labels between sites. Alternatively, collecting all data on a single server and re-training may not be feasible as data sharing agreements may restrict image transfer, or datasets may only become available at different times. Furthermore, time and cost associated with re-training grows with every new dataset. We propose a class-incremental learning method which learns an expert network for each dataset, and combines all expert networks with a score fusion model. The influence of ``unqualified experts'' is minimised by weighting each contribution with a learnt in-distribution score. These weights promote transparency as the contribution of each expert is known during inference. Instead of using the original images, we use learned features from each dataset, which are easier to share and raise fewer licensing and privacy concerns. We validate our work on six datasets from multiple sites, demonstrating significant reductions in training time while improving view classification performance.