Connecting the Dots: Where Priorities Like Mammography, Reporting and Data Systems and Artificial Intelligence Intersect

Dr.McGinty(updated)_andcroppedThis post was contributed by Geraldine McGinty, MD, MBA, FACR, chair of the ACR Board of Chancellors

We’re now more than halfway through Breast Cancer Awareness Month, and our Mammography Saves Lives campaign has been hard at work educating women and providers on the importance of annual mammograms starting at age 40 – something that we’re committed to all year long.

And if you’ve been practicing medicine for a while, you’re also likely familiar with the ACR Reporting and Data Systems (RADS), which provide a standardized framework for reporting on imaging findings with the goals of reducing the variability of terminology in reports and easing communication between radiologists and referring physicians.

Lastly, similar to Mark Twain, the rumors of radiology’s death at the hands of artificial intelligence (AI) are greatly exaggerated. As the work of the ACR Data Science Institute (ACR DSI) is demonstrating, collaboration between radiology professionals, industry leaders, government agencies, and patients is leading to the development and implementation of AI applications that will help radiology professionals improve medical care.

When viewing these efforts from a high level, it’s clear that they’re significant to radiology, but it may seem that they’re largely unrelated to one another, working in silos. But what makes an article we recently published in the JACR so exciting is seeing how these efforts converge.

The study’s authors used Breast Imaging Reporting and Data System (BI-RADS) data from the University of California, Irvine Machine Learning Repository and the Digital Database for Screening Mammography repository. Two sets of models were trained: M1 and M2. M1 used lesion shape, margin, density and patient age information from data set 1 and image texture parameters from data set 2. M2 used the same image parameters as M1, but also used BI-RADS classification provided by radiologists.

Overall, the model that used BI-RADS classification from radiologists (M2) outperformed the model that did not (M1). In simpler terms: AI algorithms perform significantly better when they include a radiologist’s opinion.

The study results demonstrate that a radiologist-augmented workflow is feasible in AI, allowing better management of patients and disease classification.

By educating women on the importance of starting mammography at age 40, leveraging the data available in BI-RADS and bringing radiologists and AI algorithms together, we can play a leading role in reducing breast cancer death rates nationwide.

  • Do you and/or your practice participate in any, or all, of the above efforts?
  • Have you heard that the ACR DSI is making a big announcement at the ACR Quality & Safety Conference next Friday, Oct. 26? What are you anticipating?

Please share your thoughts in the comments section below and join the discussion on Engage (login required).

Leave a Comment

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s