Memory-aware curriculum federated learning for breast cancer classification

Abstract

For early breast cancer detection, regular screening with mammography imaging is recommended. Routinary examinations result in datasets with a predominant amount of negative samples. A potential solution to such class-imbalance is joining forces across multiple institutions. Developing a collaborative computer-aided diagnosis system is challenging in different ways. Patient privacy and regulations need to be carefully respected. Data across institutions may be acquired from different devices or imaging protocols, leading to heterogeneous non-IID data. Also, for learning-based methods, new optimization strategies working on distributed data are required. Recently, federated learning has emerged as an effective tool for collaborative learning. In this setting, local models perform computation on their private data to update the global model. The order and the frequency of local updates influence the final global model. Hence, the order in which samples are locally presented to the optimizers plays an important role. In this work, we define a memoryaware curriculum learning method for the federated setting.Our curriculum controls the order of the training samples paying special attention to those that are forgotten after the deployment of the global model. Our approach is combined with unsupervised domain adaptation to deal with domain shift while preserving data privacy. We evaluate our method with three clinical datasets from different vendors. Our results verify the effectiveness of federated adversarial learning for the multi-site breast cancer classification. Moreover, we show that our proposed memoryaware curriculum method is beneficial to further improve classification performance. Our code is publicly available at: https://github.com/ameliajimenez/curriculum-federatedlearning.

A Review of Medical Federated Learning: Applications in Oncology and Cancer Research

“…While the bulk of the papers we've reviewed so far focus purely on designing federated algorithms that can predict different aspects of cancer with high degrees of accuracy, a large sub-group of the papers in our review also aim at addressing challenges federated learning currently faces. For many papers, that challenge is either data heterogeneity [58][59][60][61][62][63][64][65], a common barrier in the medi-cal field where patients can be subject to different geographic and demographic conditions, or label deficiency [66,67], where it is not always guaranteed that clients' sites will have access to labeled data.…”

“…Many papers modify the original FL algorithm to account for this. Jimenez et al [58] designed a novel weight aggregation algorithm designed to address the problem of domain shift between data from different institutions. This study utilized one public and two private datasets, and the final global model outperformed previous Federated Learning approaches.…”

Federated Learning: Issues in Medical Application

“…Roth [44] and Jimenez [25] used mammography dataset to classify breast density and cancer, respectively. Both results showed higher classification accuracy in federated learning than centralized learning.…”

FedBalancer: Data and Pace Control for Efficient Federated Learning on Heterogeneous Clients

“…FL trains on a large corpus of private user data without collecting them, with only the model weight updates communicated externally from the user's device [8]. With FL, researchers have proposed to improve AI in diverse domains: human mobility prediction [20], RF localization [14], traffic sign classification [4], tumour detection [28,43], and Clinical Decision Support (CDS) model for COVID-19 [16]. FL also has product deployments as large companies such as Google or Taobao deploy language processing and item recommendation tasks across millions of real-world devices [49,70].…”