This Study Looks At How AI Has Potential To Reduce The Diagnostic Divide

 The World Health Organization cites that two-thirds of the world's population do not have access to essential radiology services. Access ranges from advanced imaging like MRI and mammography but includes more basics like ultrasound and x-rays. Even a basic x-ray can show a fractured bone or lung infection.

RAD-AIG, a non-profit working to provide access to medical imaging and radiology in low-resource regions of the world, created a RadiolS-Detect is a commercially available AI that has shown diagnostic accuracy in detecting breast cancer across many studies. The Butterfly ultrasound system received US Food and Drug Administration 510(k) clearance for human use in October 2017.

Patients presenting with a palpable breast lump at the clinic at the University of Rochester were voluntarily enrolled in the study, supervised by Dr. Thomas Marini, Imaging Sciences Instructor and study lead at URMC. Patients received a Volume Sweep Image (VSI) exam with a Butterfly probe, two hours of training, and the current standard of care ultrasound exam performed by an expert breast sonographer on a a an ultrasound machine. In the study, there were 115 breast lumps tested by the AI.ogy-Readiness tool that was endorsed by the World Health Organization in 2011. The organization also offers a portfolio of artificial intelligence (AI) tools with its partners for low-resource hospitals with significant shortages of medical professionals.

But the lack of access to diagnostic imaging and services globally still puts three to four billion people at risk that could be treated if radiology services were available.

A University of Rochester Medical Center (URMC) study showed the potential of artificial intelligence (AI) to accurately diagnose breast cancer without a sonographer or radiologist. The URMC study used the Butterfly Network's whole-body portable ultrasound solution using Samsung's S-Detect AI.

In the study, the technology replaces the sonographer using VSI. "VSI involves the use of a simple ultrasound protocol based on external body landmarks, in this case, the breast lump, which is marked with an X," said Marini. "People can learn to do VSI over minutes to hours instead of months to years to train a sonographer."

Marini added that the system replaces a radiologist with Samsung's S-Detect AI. "When this all comes together, the system would enable rapid automatic diagnosis without a radiologist or a sonographer using a portable probe and a tablet to diagnose a breast lump."

Dr. Sachita Shah, Senior Director, Global Health at the Butterfly Network said the broadest impact of this technology is in diagnostic deserts. "In these environments where there is no or limited access to imaging or imaging specialists to improve referral to specialists/biopsy in a timely fashion [..] speeds definitive diagnosis and reduces time to medical and surgical oncologic care.”

"Ultrasound is portable, cost-effective, and doesn't expose patients to radiation, but its deployment is limited by a lack of specialists to acquire the images and interpret them,” said Marini. “The gold standard would always be an experienced breast sonographer certified in breast ultrasound obtaining the imaging and a dedicated breast radiologist interpreting the imaging, unfortunately, this isn’t possible for most of the world.”

“In thise setting, the alternative to VSI and AI may be no imaging at all,” adds Marini.

Marini says this proof of concept could also be applied to underserved populations in the United States.

"Removing the need to have an experienced specialist to acquire the images and removing the need for a specialist to interpret the images, significant barriers to ultrasound access are removed," said Marini. "The system we've shown proof-of-concept for would require only a tablet with AI software, a portable ultrasound probe, and an operator with a few hours of training."

Marini adds this means there are tremendous implications for global health and access to care. "This could revolutionize breast cancer diagnosis in low- and middle-income countries [..] where significant delays in breast cancer diagnoses contribute to advanced disease and death.”