UB-affiliated company uses AI to assist surgeons treating brain aneurysms in real time

Brain, or intracranial aneurysms (IA), can lead to hemorrhagic strokes and are responsible for close to 500,000 deaths worldwide every year.

Despite advances in medical treatments, around 65% of patients with ruptured IAs die from the initial bleed or subsequent complications, notes Ciprian “Chip” Ionita, assistant professor of biomedical engineering and neurosurgery. Among the survivors, about 50% become disabled with significant loss of independence.

Ionita, who also serves as director of the Endovascular Devices and Imaging Lab at Canon Stroke and Vascular Research Center (CSVRC), hopes to improve these odds through Quantitative Angiographic Systems. Artificial Intelligence (QAS.AI), a small biomedical engineering company he founded at the CSVRC in 2020.

Ionita and his research team are developing software intended to detect complications during surgery, such as inadequate blood flow in the brain, and forecast whether a selected treatment method will succeed — all in real time.

“The technology developed by QAS is fundamentally different than current FDA-approved AI solutions, which are used mostly for offline diagnosis or clinical workflow optimization,” Ionita explains. “We are pushing the boundary of the AI applications, developing prognosis tools for the operating room, which could improve treatment outcomes and early detection of complications. This will cut health care costs and save lives.”

Now, Ionita and his team have the opportunity to apply their research to clinical evaluations in two sites in Buffalo and one in Florida, thanks to a $1 million, Phase II grant from the National Science Foundation’s (NSF) Small Business Technology Transfer (STTR) program.

Located at UB’s New York State Center of Excellence in Bioinformatics and Life Sciences, QAS.IA integrates years of research by faculty and students in the Jacobs School of Medicine and Biomedical Sciences and the School of Engineering and Applied Sciences.

In 2021 QAS.AI received $256,000 in initial funding from the same NSF program to begin creating the AI-based software.

“The AI software we developed is integral during the intervention process, as it assesses the likelihood of aneurysm healing,” he said. “If it predicts a low probability of healing within a year, this information is immediately relayed to the neurosurgeons. This enables them to consider adjusting the treatment approach, potentially by incorporating an additional device. This feature is crucial for keeping a close watch on the patient’s condition, helping doctors respond effectively to changes in the aneurysm’s behavior.”

The new grant, which extends from fall 2023 to fall 2025, will fund the hiring of a company to develop clinical-grade software to translate the current software that QAS.AI is using to accomplish these goals.

“It will ensure that our software is HIPPA [Health Insurance Portability and Accountability Act] compliant, has patient protection and all the robustness required by clinical software,” Ionita said. “It has to be completely integrated with surgical equipment.”

The grant is also funding clinical evaluations at the Gates Vascular Institute (GVI), Mercy Hospital of Buffalo and the University of South Florida’s Department of Neurosurgery and Brain Repair.

Ionita, who conceived the AI software, and is the QAS-AI chief executive officer, has engaged biomedical engineering students in a variety of projects to advance the technology. Mohammad Mahdi Shiraz Bhurwani, who served as the company’s lead AI scientist in summer 2021 while completing his doctoral research in the department, is now the grant’s principal investigator and the company’s chief technical officer.

Other team members include Jason Davies, assistant professor of neurosurgery and biomedical informatics, who serves as chief medical officer; Vincent Tutino, assistant professor of pathology and anatomical sciences, who serves as the chief financial officer; and two PhD candidates.

The platform for Phase II includes a fully automated method to identify the location and extent of the IA. It also allows for instantaneous extraction of the imaging biomarkers, prognosis of the surgical outcome at one-year post-procedure in a fraction of a second to allow neuro-interventionalist to readjust the endovascular therapy, and full integration with the angiographic systems regardless of the manufacturer, Ionita explained.

“When brain aneurysms become symptomatic and traditional treatments such as opening the skull prove ineffective, less invasive neuro-endovascular interventions can offer a solution,” Ionita said. “This can be accomplished by placing coils, stents or a combination of these devices in the region of an aneurysm.”

However, he said, they don’t always result in complete IA healing. In fact, between 70 to 80% of these AI cases are treated successfully while the other 20 to 30% require the patient to come back for additional procedures.

“We want to reduce this number of re-treatments to single digits, if possible,” Ionita said. “Our goal is to develop a technology that allows successful treatment in one stop.” 

Ionita expects the first clinical evaluation at GVI to be operational by August 2024; evaluations at the other two spots will follow.

“Through the clinical evaluations, we will be preparing for clinical trials,” he said, adding that they will require future grants or private investments but should lead to FDA approval of the AI software.

The potential commercial impact of this technology is immense, Ionita said, especially at a time when medical imaging companies are seeking new innovations to maintain a competitive edge.

“By integrating an intraoperative software for informed decision-making, these companies could experience a 5% increase in scanner sales,” he said. “With over 6,000 hospitals in the U.S., each equipped with an average of six angiographic suites, this translates to projected revenue of $1.1 billion in U.S. markets.”

In addition, hospital and insurance company administrators will benefit from reduced re-treatment costs, which average $65,000 each and would amount to an annual savings of $1.95 billion in the United States alone, he said.

“This innovative AI platform promises to revolutionize the medical imaging and health care industries,” he said, “improving patient outcomes while boosting the bottom line.”