Back to feed
arXiv cs.LG
arXiv cs.LG
7/22/2026
FedCC: A Low-Resource Federated Adaptation of Foundation Models for Robust Corpus Callosum localization in Fetal Ultrasound Images

FedCC: A Low-Resource Federated Adaptation of Foundation Models for Robust Corpus Callosum localization in Fetal Ultrasound Images

Short summary

FedCC is a federated learning framework for localizing the corpus callosum in fetal ultrasound images across multi-center, resource-constrained clinical settings without data sharing. It combines a frozen DINOv2 backbone with a lightweight YOLO detection head and LoRA modules, reducing trainable parameters from 24.4M to 2.9M (8.5x communication cost reduction). Evaluated on 10,970 frames from 58 pregnant women across three sites, the DINOv2+LoRA FedAvg configuration achieved 0.857 mAP@50 and 0.803 F1-score, outperforming full fine-tuning and encoder-freezing baselines.

  • •Federated learning framework for fetal ultrasound corpus callosum localization without data sharing
  • •Frozen DINOv2 + YOLO + LoRA reduces trainable params 8.5x to 2.9M while achieving 0.857 mAP@50
  • •Validated on multi-center dataset of 10,970 frames from 58 patients across three clinical sites

Generated with AI, which can make mistakes.

#research-breakthrough#open-source
Read full article at arXiv cs.LG

Is this a good recommendation for you?

Comments

Failed to load comments. Please try again.

Explore more

Federated Learning for Object Detection: Enabling Collaborative Drone Learning Without Centralizing Data
arXiv cs.LG
HantaWatch: Federated Learning for Hantavirus Genomic Surveillance
arXiv cs.LG
SUM: Unified Geometric Surgery on Spatio-Temporal Adaptation Vectors for Federated Class Incremental Learning
arXiv cs.LG
WHERE to Generate Matters: Budget-Aware Synthetic Augmentation for Label Skewed Federated Learning
arXiv cs.LG