Curriculum Learning for Few-Shot Domain Adaptation in CT-based Airway Tree Segmentation
This work addresses the challenge of expensive manual annotation in medical imaging by enabling few-shot domain adaptation for airway tree segmentation, though it is incremental as it builds on existing Curriculum Learning methods.
The authors tackled the problem of poor generalization in automated airway segmentation from chest CT scans by integrating Curriculum Learning into deep learning networks, achieving high performance on two large open-cohorts (ATM22 and AIIB23) for both full training and few-shot fine-tuning.
Despite advances with deep learning (DL), automated airway segmentation from chest CT scans continues to face challenges in segmentation quality and generalization across cohorts. To address these, we propose integrating Curriculum Learning (CL) into airway segmentation networks, distributing the training set into batches according to ad-hoc complexity scores derived from CT scans and corresponding ground-truth tree features. We specifically investigate few-shot domain adaptation, targeting scenarios where manual annotation of a full fine-tuning dataset is prohibitively expensive. Results are reported on two large open-cohorts (ATM22 and AIIB23) with high performance using CL for full training (Source domain) and few-shot fine-tuning (Target domain), but with also some insights on potential detrimental effects if using a classic Bootstrapping scoring function or if not using proper scan sequencing.