Abstract: Although hospitals have accumulated a tremendous number of radiology images and reports, using them to build data-hungry deep-learning models remains challenging. This is partially because manually annotating a dataset in large scale is costly. Here, we propose using text mining to automatically generate a weakly-labeled dataset for detecting thoracic diseases in radiology images. This work represents the first attempt to unify text mining with computer vision for medical imaging analysis in the era of deep learning.

Learning Objective 1: After participating in this session, the learner should be better able to learn and evaluate our unified framework, involving: our framework workflow, how to automatically generate labels for X-ray reports, and how to use our system to detect thorax diseases in X-ray images.


Yifan Peng (Presenter)
National Institutes of Health

Xiaosong Wang, National Institutes of Health
Le Lu, National Institutes of Health
Bagheri Mohammadhadi, National Institutes of Health
Ronald Summers, National Institutes of Health
Zhiyong Lu, National Institutes of Health

Presentation Materials: