Effect of Radiology Report Labeler Quality on Deep Learning Models for Chest X-Ray Interpretation
This work addresses the problem of improving chest X-ray classification models for medical diagnosis by showing that better labeling tools can enhance model accuracy, though it is incremental as it builds on existing labelers.
The study investigated how the quality of radiology report labelers affects deep learning models for chest X-ray interpretation, finding that using the VisualCheXbert labeler led to better model performance compared to CheXpert and CheXbert labelers.
Although deep learning models for chest X-ray interpretation are commonly trained on labels generated by automatic radiology report labelers, the impact of improvements in report labeling on the performance of chest X-ray classification models has not been systematically investigated. We first compare the CheXpert, CheXbert, and VisualCheXbert labelers on the task of extracting accurate chest X-ray image labels from radiology reports, reporting that the VisualCheXbert labeler outperforms the CheXpert and CheXbert labelers. Next, after training image classification models using labels generated from the different radiology report labelers on one of the largest datasets of chest X-rays, we show that an image classification model trained on labels from the VisualCheXbert labeler outperforms image classification models trained on labels from the CheXpert and CheXbert labelers. Our work suggests that recent improvements in radiology report labeling can translate to the development of higher performing chest X-ray classification models.