Unlocking Robust Segmentation Across All Age Groups via Continual Learning

arXiv:2404.13185v1 Announce Type: cross
Abstract: Most deep learning models in medical imaging are trained on adult data with unclear performance on pediatric images. In this work, we aim to address this challenge in the context of automated anatomy segmentation in whole-body Computed Tomography (CT). We evaluate the performance of CT organ segmentation algorithms trained on adult data when applied to pediatric CT volumes and identify substantial age-dependent underperformance. We subsequently propose and evaluate strategies, including data augmentation and continual learning …

abstract age aim arxiv automated challenge context continual cs.cv data deep learning eess.iv images imaging medical medical imaging performance robust segmentation type via work