Northeast Ohio hospitals are turning to artificial intelligence to provide patients with faster, more accurate results after they undergo CT scans, mammograms, MRIs and other tests.
Summa Health, University Hospitals and Cleveland Clinic all have established usage of AI in radiology in recent years to aid — but not replace — physicians with determining diagnoses and care plans.
The technology can benefit patients in multiple ways. It can pick up abnormalities in scans quickly, leading to receiving treatment for potentially life-threatening diagnoses faster. It increases result accuracy when doctors use it as a second pair of eyes, and it reduces the amount of time patients have to spend in an uncomfortable MRI machine with faster image reconstruction.
“No AI radiology is running in a way that it is superimposing or overlapping or replacing the radiologist’s judgment,” said Dr. Leonardo Kayat Bittencourt, vice chair of innovation for University Hospitals. “None of them, absolutely none of them, runs autonomously in the sense that it speeds out the result that is not criticized, is not visualized, is not adjudicated by a radiologist.”
“I’ll call it a connective tissue that brings our services together,” said Dr. Po-Hao Chen, vice chair for artificial intelligence in the Diagnostics Institute at the Cleveland Clinic. “That AI is never making a diagnosis on its own. Every time that it touches a patient’s care, it’s always been proctored or overseen by the human.”
How hospitals are using AI in radiology
Each hospital uses AI to aid radiologists in flagging suspicious areas on scans and identifying the highest-risk patients who need to be seen.
At Summa Health, Dr. Brian Bauman, head of pulmonary services, said AI is used to aid radiologists in detecting lung nodules, which are small masses of tissue in a lung that can be cancerous but are often benign.
“Essentially, what those models do is they can actually analyze images and then flag certain findings for the radiologist to pick up more easily,” said Bauman, who is also a pulmonary and critical care physician.
When a radiologist sees a potentially cancerous lung nodule, which is life-threatening, they can take action to treat it more quickly, Bauman said.
Then, instead of a lung nodule navigator nurse going through thousands of reports, AI algorithms create a list of which patients are at higher risk and need to be seen quicker, Bauman said.
This natural language software picks up on keywords like “nodule” or “density.” These lists are then sent to the nurses, who can take it from there.
“Just finding these things isn’t necessarily helpful if you don’t have a process to handle the findings once they’re identified,” he said.
Both University Hospitals and Summa Health use ClearRead CT, owned by Riverain Technologies, to aid in interpreting CT scan results. The technology was approved by the Food and Drug Administration in 2016. Studies show it can improve detection of previously missed nodules by 29% and reduce the time it takes to interpret results by 36%.
“Ultimately, it’s an additional layer of safety and health that those patients are getting on top of all the expertise of our team,” said Bittencourt, who also works in abdominal imaging.
Chen, who is also a diagnostic radiologist, said the Cleveland Clinic also uses AI to accelerate image reconstruction from MRIs. The Cleveland Clinic system of hospitals includes Akron General, Medina Hospital and Mercy Hospital in Canton.
Nobody likes being in an MRI machine, he said. The algorithm is used to substantially cut down the amount of time a patient has to spend inside the scanner.
Benefits of using AI in radiology
Bauman said that, oftentimes, lung nodules are found on CT scans that were originally performed for other reasons. Because of this, they frequently get missed because it’s not what the radiologist is looking for.
In 2025, about 3,400 patients at Summa Health were found to have lung nodules, and 357 were identified by accident as having cancer, meaning they had no symptoms and had a CT scan for another reason.
“If you catch those nodules early, then you can find lung cancer early, and then that actually improves people’s outcomes dramatically,” he said.
AI can also review prior CT scans, while a radiologist checks current ones, to see if the nodule grew, Bauman said. This reduces the amount of time and effort it would take for the radiologist to look at the prior exams.
In mammography, Chen said algorithms can recommend taking a closer look at a mass in a breast by outlining where it is. It can also increase diagnostic certainty.
“In some cases, AI can serve as a sanity check, or as a second pair of eyes, and basically help the expert lean one way or another,” Chen said.
Because a CT scan can have a thousand images, he said a human expert takes a lot of time trying to determine whether something is suspicious. With AI, if a doctor is already concerned about a specific area, the AI can confirm their suspicion, increasing a doctor’s certainty of their finding.
Reports also are developed faster because AI can write a concise conclusion that the radiologist can then edit for accuracy, Chen said. This cuts down on drafting redundancy, as doctors have to painstakingly go through everything that could be wrong in a scan.
“That’s a lot faster than trying to dictate each of those one by one,” Chen said.
Potential consequences of using AI in radiology
AI in hospitals is a work in progress, Bauman said.
“The really important part about AI is making sure that it’s combined with human intelligence,” Bauman said. “We don’t want to purely rely on the AI.”
Chen said AI can be wrong, but the mistakes it makes are different from a human expert.
“When you put two and two together, the human can spot many of the AI errors quickly, and, likewise, AI can spot some of the human errors quickly,” he said.
Because there are several moving pieces in the background of a running algorithm, with data being moved across a network, it always needs to be vetted and validated, Chen said.
He gave a hypothetical example of the internet going down in one of the Cleveland Clinic’s data centers. A radiologist will still get a scan, but the AI results may not come back as fast as they normally do.
“If your workflow has come to rely on an AI result, and that just becomes your new normal, when it comes back a little bit later, then it might take a few more minutes for you to generate the rest of your work,” he said.
Do hospitals disclose AI use to patients?
Bauman and Chen said that, generally, patients are not explicitly told when AI is used behind the scenes in their health care.
“I think the consensus is that this is becoming standard of care and that this isn’t replacing anything that would have already been done,” Bauman said. “It’s just helping the providers who are already taking care of that patient do a better job.”
Chen said mammography does tell patients that a computer-aided detection tool was used in the interpretation of their case, and the department has always done so.
Bittencourt said that, at University Hospitals, the use of AI is always human-centered.
“Nobody’s thinking about AI as a means to cut costs or to replace the human expertise, or to take humans from the loop, but much more so to enhance us humans, make us better and make us better doctors to our patients,” he said.
Lauren Cohen is a community reporting intern for the Akron Beacon Journal and Signal Akron. The position is funded through a grant from the Knight Foundation, which is also a financial supporter of Signal Akron.