Lung-RADS Assist: Artificial Intelligence Model Verification, Reporting, and Monitoring

  • Overview

    The American College of Radiology Lung CT Screening Reporting and Data System (Lung-RADS) is designed to standardize lung cancer screening and provide management recommendations. The Lung Cancer Screening Registry (LCSR) allows data to be captured for quality reporting as required by CMS for lung cancer screening payment. Designed as an artificial intelligence (AI) validation study and artificial intelligence surveillance model, this project evaluates pre-market model validation and post-market model performance for an artificial intelligence algorithm used for lung cancer screening.

    Specific aims include the following:

    1. Utilize existing ACR technology to demonstrate the ability to collect validation data and perform local algorithm testing prior to market approval.
    2. Utilize existing ACR technology to facilitate interoperability between reporting and AI vendors to generate standardized data in a real world setting.
    3. Capture validation data and real world events in a national registry to enable both facility-level and cross-facility reporting.
  • Impact for NESTcc

    The project allows for the testing and validation of an artificial intelligence algorithm. It will provide learnings that will be scalable to other disease areas. It will include the capture of performance metrics within a national registry linking both AI model validation and AI model surveillance, providing real world evidence to measure safety and effectiveness.

  • Organizations

    American College of Radiology

    • Implementation Sites include: Brigham & Women’s, Massachusetts General Hospital
    • Radiology Workflow Companies include: GE Healthcare, Nuance
    • Algorithm Vendors include: Aidence, Enlitic, Infervision, Mindshare Medical
  • Principal Investigators

    Keith Dreyer, DO, PhD, Massachusetts General Hospital