CASED: Curriculum Adaptive Sampling for Extreme Data Imbalance

TL;DR

The paper presents CASED, a curriculum sampling algorithm that significantly improves deep learning segmentation models on highly imbalanced datasets, achieving state-of-the-art results in lung nodule detection with minimal detection stages.

Contribution

Introduces CASED, a novel curriculum sampling method that enhances training of deep segmentation models on imbalanced data, outperforming traditional two-stage detection approaches.

Findings

Achieves 88.35% sensitivity on LUNA16 benchmark.

Models trained with CASED are robust to annotation quality.

Framework generalizes across imaging modalities and segmentation targets.

Abstract

We introduce CASED, a novel curriculum sampling algorithm that facilitates the optimization of deep learning segmentation or detection models on data sets with extreme class imbalance. We evaluate the CASED learning framework on the task of lung nodule detection in chest CT. In contrast to two-stage solutions, wherein nodule candidates are first proposed by a segmentation model and refined by a second detection stage, CASED improves the training of deep nodule segmentation models (e.g. UNet) to the point where state of the art results are achieved using only a trivial detection stage. CASED improves the optimization of deep segmentation models by allowing them to first learn how to distinguish nodules from their immediate surroundings, while continuously adding a greater proportion of difficult-to-classify global context, until uniformly sampling from the empirical data distribution. Using CASED during training yields a minimalist proposal to the lung nodule detection problem that tops the LUNA16 nodule detection benchmark with an average sensitivity score of 88.35%. Furthermore, we find that models trained using CASED are robust to nodule annotation quality by showing that comparable results can be achieved when only a point and radius for each ground truth nodule are provided during training. Finally, the CASED learning framework makes no assumptions with regard to imaging modality or segmentation target and should generalize to other medical imaging problems where class imbalance is a persistent problem.