Curriculum Learning for Few-Shot Domain Adaptation in CT-based Airway Tree Segmentation

TL;DR

This paper introduces a curriculum learning approach to improve airway segmentation in chest CT scans, especially in few-shot domain adaptation scenarios, by organizing training data based on complexity scores.

Contribution

It presents a novel integration of curriculum learning into airway segmentation networks for better generalization and effective few-shot domain adaptation in medical imaging.

Findings

High performance on large cohorts with curriculum learning

Insights into potential pitfalls of classic scoring functions

Effective few-shot fine-tuning for domain adaptation

Abstract

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.