cnythzl/DigitalVision Vectors vi
Geisinger Deploys AI Imaging Model to Diagnose Coronary Artery Disease
Clinicians at Geisinger have adopted a non-invasive medical imaging model that uses deep learning to create 3D models of a patient’s arteries, which can help diagnose coronary artery disease.
Clinicians at Pennsylvania-based Geisinger health system have deployed an artificial intelligence (AI) medical imaging model that uses computed tomography (CT) scans to create 3D models of a patient's arteries, helping to diagnose coronary artery disease (CAD) without the use of invasive procedures.
Called HeartFlow Analysis, the AI imaging model involves Geisinger clinicians taking data from a patient’s coronary CT scans to generate 3D models of their coronary arteries, allowing clinicians to assess a patient for CAD without performing unnecessary invasive procedures in the operating room or tests in the emergency department.
“HeartFlow Analysis uses computer algorithms to simulate blood flow, so we can assess the impact CAD has on blood flow to the heart with a non-invasive test,” said Vishal Mehra, MD, Geisinger’s chief of advanced cardiac imaging, in the press release. “This enhances our ability to diagnose and treat patients for whom cardiac catheterization is not appropriate or necessary and to avoid delays for patients who need cardiac catheterization urgently.”
CAD is one of the most common heart conditions and the leading cause of death among Americans. CAD occurs when cholesterol plaques develop and accumulate in the coronary arteries, obstructing the flow of oxygen-rich blood to the heart. This obstructed blood flow can lead to disabling symptoms, such as shortness of breath and chest pain, or more severe outcomes like heart attacks and death.
CAD is typically diagnosed using cardiac catheterization, which is used to measure the fractional flow reserve (FFR). FFR measures the maximum amount of blood that can flow through a diseased, narrow artery compared to how much blood can flow through a healthy artery, which shows if delivery of oxygen to the heart has been compromised.
Cardiac catheterization is an invasive procedure, and any invasive procedure can result in complications. But it has historically been one of the only ways to diagnose CAD. As a result, unnecessary cardiac catheterization and other tests have been performed on patients just to rule out CAD. Thus, Geisinger turned to HeartFlow Analysis.
“Embracing innovative technology helps us put our patients first,” said George Ruiz, MD, chair of Geisinger’s department of cardiology, in the press release. “With HeartFlow Analysis, no sedation or overnight stay is necessary, and we can obtain critical information easily.”
Geisinger is currently the only health system utilizing HeartFlow Analysis, also known as fractional flow reserve – computed tomography (FFR-CT). The technology is available at all Geisinger hospitals where cardiac CT is performed and is part of the health system’s strategy to enhance cardiac CT.
The adoption of HeartFlow Analysis builds on Geisinger’s other recent efforts to improve heart disease diagnosis.
Earlier this week, Geisinger and biotechnology research company Tempus announced that they had developed an AI model to predict undiagnosed structural heart disease (SHD). The model, known as rECHOmmend, combines electrocardiogram (ECG) data with a deep neural network to predict which patients, among those without a prior history of SHD, would develop the disease and could benefit from monitoring or treatment.
The model achieved high performance at predicting the risk of developing any one of seven SHDs that are diagnosable via a standard ECG. It also significantly outperformed previously published models designed to predict any single SHD.
These findings indicate that clinicians using rECHOmmend could identify more diseases using fewer diagnostic studies, which could mean fewer tests for patients.