Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Recommender Engine Driven Client Selection in Federated Brain Tumor Segmentation

Muhammad Irfan Khan, Elina Kontio, Suleiman A. Khan, Mojtaba Jafaritadi

TL;DR

This work tackles efficient collaborator selection in Federated Learning for brain tumor segmentation under non-IID data and resource heterogeneity. It introduces a NNMF-based recommender engine to dynamically select 20% of collaborators per round and a Harmonic Similarity Weighted Aggregation (HSimAgg) to robustly aggregate model updates. Evaluated on FeTS-derived mpMRI data with a 3D U-Net, the approach achieves external validation Dice scores of ET 0.7298, TC 0.7424, and WT 0.8218, with rapid convergence. The findings demonstrate that task-aligned collaborator selection and harmonic-weighted aggregation can improve FL precision and efficiency, offering practical benefits for privacy-preserving medical image analysis.

Abstract

This study presents a robust and efficient client selection protocol designed to optimize the Federated Learning (FL) process for the Federated Tumor Segmentation Challenge (FeTS 2024). In the evolving landscape of FL, the judicious selection of collaborators emerges as a critical determinant for the success and efficiency of collective learning endeavors, particularly in domains requiring high precision. This work introduces a recommender engine framework based on non-negative matrix factorization (NNMF) and a hybrid aggregation approach that blends content-based and collaborative filtering. This method intelligently analyzes historical performance, expertise, and other relevant metrics to identify the most suitable collaborators. This approach not only addresses the cold start problem where new or inactive collaborators pose selection challenges due to limited data but also significantly improves the precision and efficiency of the FL process. Additionally, we propose harmonic similarity weight aggregation (HSimAgg) for adaptive aggregation of model parameters. We utilized a dataset comprising 1,251 multi-parametric magnetic resonance imaging (mpMRI) scans from individuals diagnosed with glioblastoma (GBM) for training purposes and an additional 219 mpMRI scans for external evaluations. Our federated tumor segmentation approach achieved dice scores of 0.7298, 0.7424, and 0.8218 for enhancing tumor (ET), tumor core (TC), and whole tumor (WT) segmentation tasks respectively on the external validation set. In conclusion, this research demonstrates that selecting collaborators with expertise aligned to specific tasks, like brain tumor segmentation, improves the effectiveness of FL networks.

Recommender Engine Driven Client Selection in Federated Brain Tumor Segmentation

TL;DR

This work tackles efficient collaborator selection in Federated Learning for brain tumor segmentation under non-IID data and resource heterogeneity. It introduces a NNMF-based recommender engine to dynamically select 20% of collaborators per round and a Harmonic Similarity Weighted Aggregation (HSimAgg) to robustly aggregate model updates. Evaluated on FeTS-derived mpMRI data with a 3D U-Net, the approach achieves external validation Dice scores of ET 0.7298, TC 0.7424, and WT 0.8218, with rapid convergence. The findings demonstrate that task-aligned collaborator selection and harmonic-weighted aggregation can improve FL precision and efficiency, offering practical benefits for privacy-preserving medical image analysis.

Abstract

This study presents a robust and efficient client selection protocol designed to optimize the Federated Learning (FL) process for the Federated Tumor Segmentation Challenge (FeTS 2024). In the evolving landscape of FL, the judicious selection of collaborators emerges as a critical determinant for the success and efficiency of collective learning endeavors, particularly in domains requiring high precision. This work introduces a recommender engine framework based on non-negative matrix factorization (NNMF) and a hybrid aggregation approach that blends content-based and collaborative filtering. This method intelligently analyzes historical performance, expertise, and other relevant metrics to identify the most suitable collaborators. This approach not only addresses the cold start problem where new or inactive collaborators pose selection challenges due to limited data but also significantly improves the precision and efficiency of the FL process. Additionally, we propose harmonic similarity weight aggregation (HSimAgg) for adaptive aggregation of model parameters. We utilized a dataset comprising 1,251 multi-parametric magnetic resonance imaging (mpMRI) scans from individuals diagnosed with glioblastoma (GBM) for training purposes and an additional 219 mpMRI scans for external evaluations. Our federated tumor segmentation approach achieved dice scores of 0.7298, 0.7424, and 0.8218 for enhancing tumor (ET), tumor core (TC), and whole tumor (WT) segmentation tasks respectively on the external validation set. In conclusion, this research demonstrates that selecting collaborators with expertise aligned to specific tasks, like brain tumor segmentation, improves the effectiveness of FL networks.
Paper Structure (13 sections, 6 equations, 2 figures, 3 tables, 1 algorithm)

This paper contains 13 sections, 6 equations, 2 figures, 3 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Methods
  3. Data
  4. Federated Learning Framework
  5. Recommender Engine Collaborator Selection
  6. Harmonic Similarity Weighted Aggregation (HSimAgg)
  7. Experiments
  8. Setup
  9. Results
  10. Model training and performance using internal validation data.
  11. Model performance using external validation data.
  12. Discussion
  13. Conclusion

Figures (2)

  • Figure 1: Non-negative matrix factorization (NNMF) strategy.
  • Figure 2: Performance metrics for model training of HSimAgg. The horizontal axis refers to the number of rounds and the vertical axis to the performance metrics.