Future plans call for improving accuracy of detection via the use of 3D and lesion type information, expanding the data set to include data from other institutions, and also to enlarge the data set with MR images.

The NIH Clinical Center made news in 2017 when it released more than 100,000 anonymized chest x-rays and corresponding data “to allow researchers across the country and around the world to freely access the datasets and increase their ability to teach computers how to detect and diagnose disease,” it said at that time.


That data set, known as “ChestX-ray8”, came from scans on over 30,000 patients and was also released by Summers and an NIH team.

“Building truly large-scale, fully automated, high-precision medical diagnosis systems remains a strenuous task,” stated lead author Xiaosong Wang and senior author Summers in the paper presenting the data set. They noted that “ChestX-ray8” can enable the data-hungry deep neural network paradigms to create clinically meaningful applications, including common disease pattern mining, disease correlation analysis and automated radiological report generation.”

The DeepLesion database can be downloaded at https://nihcc.box.com/v/DeepLesion.

Back to HCB News

CT Homepage