URL: Combating Label Noise for Lung Nodule Malignancy Grading

TL;DR

This paper introduces the URL framework, which improves lung nodule malignancy grading by modeling class ordinal relations and handling label noise through contrastive learning, pseudo-labels, and unimodal regularization.

Contribution

It proposes a novel two-stage URL framework that incorporates ordinal relations and pseudo-labeling to enhance robustness against label noise in lung nodule grading.

Findings

URL outperforms existing methods on LIDC-IDRI dataset

The framework effectively models ordinal class relations

Pseudo-labeling improves label noise robustness

Abstract

Due to the complexity of annotation and inter-annotator variability, most lung nodule malignancy grading datasets contain label noise, which inevitably degrades the performance and generalizability of models. Although researchers adopt the label-noise-robust methods to handle label noise for lung nodule malignancy grading, they do not consider the inherent ordinal relation among classes of this task. To model the ordinal relation among classes to facilitate tackling label noise in this task, we propose a Unimodal-Regularized Label-noise-tolerant (URL) framework. Our URL contains two stages, the Supervised Contrastive Learning (SCL) stage and the Memory pseudo-labels generation and Unimodal regularization (MU) stage. In the SCL stage, we select reliable samples and adopt supervised contrastive learning to learn better representations. In the MU stage, we split samples with multiple annotations into multiple samples with a single annotation and shuffle them into different batches. To handle label noise, pseudo-labels are generated using the similarity between each sample and the central feature of each class, and temporal ensembling is used to obtain memory pseudo-labels that supervise the model training. To model the ordinal relation, we introduce unimodal regularization to keep the ordinal relation among classes in the predictions. Moreover, each lung nodule is characterized by three orthographic views. Experiments conducted on the LIDC-IDRI dataset indicate the superiority of our URL over other competing methods. Code is available at https://github.com/axz520/UR.