Artificial intelligence? Check. University Hospitals radiologists get help from new tool

Dr. Jennifer Sommer, neuroradiologist and vice chair of informatics in the Department of Radiology at University Hospitals in Cleveland, discusses the use of artificial intelligence in radiology.
Dr. Jennifer Sommer, neuroradiologist and vice chair of informatics in the Department of Radiology at University Hospitals in Cleveland, discusses the use of artificial intelligence in radiology.

At University Hospitals in Cleveland, radiologists sit in a dark room analyzing CT scans to diagnose and rule out various patient diseases and conditions.

Earlier this year, the entire hospital system started implementing new artificial intelligence tools across radiology to look at possible health conditions and to help radiologists prioritize the order in which they first review new scans.

Drs. Jennifer Sommer and Leonardo Kayat Bittencourt, both UH radiologists, showed the Beacon Journal Aidoc's AI technology. In one patient scan, they showed that Aidoc’s AI detected a possible aortic dissection, or tear to the aorta, the body’s main artery, but that it didn’t find evidence of a pulmonary embolism — clogging of the lungs usually caused by blood clots.

These technologies add to UH’s existing use of AI, such as through a partnership with GE HealthCare to look for collapsed lungs through X-rays and linking up with AZmed to use X-rays in the search for orthopedic fractures, said Dr. Donna Plecha, UH radiology chair. With the newer Aidoc partnership, Plecha said the percentage of cases that UH radiologists is evaluating with AI is “much larger than before.”

The UH hospital system includes UH Portage Medical Center in Ravenna.

University Hospitals radiologists see benefits from using AI

Plecha said AI tools such as the ones UH is now using with Aidoc do not make diagnoses on their own, and they help rather than replace radiologists.

“And most studies that look at artificial intelligence, specifically in radiology, show that artificial intelligence with a radiologist does better than either one alone,” Plecha said.

Radiologists work on multiple computer screens at once to pull up patient information and review their images, Plecha said.

Plecha said Aidoc’s AI algorithms “will actually, when you open up the image, highlight areas that are worrisome or might be worrisome to make sure that certain things aren't missed. And then the radiologist has to determine if that's an actual, positive finding.”

False positives and false negatives can occur with the technology, Plecha said.

Bittencourt, abdominal radiologist and vice chair of innovation in radiology at UH, said he has not firsthand seen false findings and the UH radiology team has found they’re uncommon. He said Aidoc's AI technology serves as a “safety net” and said it shows radiologists with a heat map how it flags and rules out possible conditions.

Sommer, a neuroradiologist and vice chair of informatics in UH’s radiology department, said the hospital system will roll out more of Aidoc’s technology soon. One benefit of the technology, she said, is that it can detect possible brain lesions in patients with multiple sclerosis.

A monitor shows the use the artificial intelligence-driven Aidoc program that University Hospitals' radiologists implemented earlier this year  to improve interpretation of the scans in Cleveland.
A monitor shows the use the artificial intelligence-driven Aidoc program that University Hospitals' radiologists implemented earlier this year to improve interpretation of the scans in Cleveland.

Aidoc CEO says AI addresses healthcare challenges brought on by labor shortage

Elad Walach, CEO of Aidoc, said the company's technology will have uses outside of radiology, such as cardiovascular care and neuroscience.

He said the technology can speed up the time between when patients undergo testing and receive their results, adding that generally in healthcare, there can be long delays. But he said he isn’t aware of any such delays at UH, specifically.

The health care industry is facing a labor shortage right now, he said.

The radiology field specifically recognizes a labor shortage in the specialization, according to the American College of Radiology.

“So, I think we really have to rethink healthcare productivity,” Walach said, calling AI a "massive opportunity" to improve efficiencies in the absence of more health care professionals.

Of Aidoc’s AI algorithms, 17 have been cleared by the FDA, Walach said.

Walach and Plecha said UH will incrementally start using more of the algorithms until the hospital system has incorporated all 17.

Addressing the concern of false results, Walach said Aidoc implements multiple technologies in tandem at UH and other medical institutions to prevent data “drift” that occurs with other AI technologies such as ChatGPT.

Dr. Leonardo Kayat Bittencourt, left, abdominal radiologist and vice chair of innovation in University Hospitals' Department of Radiology and Dr. Jennifer Sommer, neuroradiologist and vice chair of informatics in the department, discuss the use of artificial intelligence in their field Friday in Cleveland.
Dr. Leonardo Kayat Bittencourt, left, abdominal radiologist and vice chair of innovation in University Hospitals' Department of Radiology and Dr. Jennifer Sommer, neuroradiologist and vice chair of informatics in the department, discuss the use of artificial intelligence in their field Friday in Cleveland.

Feedback on AI technology is key, collaborators say

Walach said UH importantly can provide feedback on the technology’s findings in real time.

UH has not reviewed all of the data from Aidoc yet, Plecha said, adding that the team will be more closely reviewing false findings.

“The other thing is that it has been very helpful in finding some subtle areas and subtle findings that may have otherwise not have been noticed right away,” Plecha said. “There's feedback on both the pros and the cons, but I think overall, people are very pleased with it, and I don't think we can ignore AI. I think we have to incorporate AI into our future because I think we have to take advantage of the technology but do it in a very responsible way.”

Patrick Williams covers growth and development for the Akron Beacon Journal. He can be reached by email at pwilliams@gannett.com or on X, formerly known as Twitter, @pwilliamsOH.

This article originally appeared on Akron Beacon Journal: University Hospitals using AI tool to assist with radiology diagnoses

Advertisement