An AI algorithm used on a CT image of the abdomen shows body composition that can reveal emerging health problems. Green indicates fat below the skin; red is deeper “bad fat”; blue is muscle; pink is the vertebrae; yellow is fat between muscles; and turquoise is the bowel and cerebrospinal fluid.
Courtesy of Dr. Miriam Bredella
Every year, more than 80 million computerized tomography (CT) scans are performed in the United States. These diagnostic exams use X-ray technology to generate detailed anatomical images, enabling clinicians to identify tumors, fractures, and other abnormalities.
NYU Langone Health’s Department of Radiology is convinced that the mountain of valuable data contained in these images remains largely untapped. A team of faculty, led by Miriam A. Bredella, MD, MBA, the Bernard and Irene Schwartz Professor of Radiology and director of the Clinical and Translational Science Institute; Soterios Gyftopoulos, MD, MBA, MSc, chief of radiology at NYU Langone Hospital—Brooklyn; and Bari Dane, MD, the department’s director of computerized tomography, uses information from “discarded” clinical CTs to uncover emerging health problems, a process called opportunistic imaging.
Advances in both artificial intelligence, or AI, and machine learning have allowed radiologists to quantify the amount of fat, muscle, mineral deposits in veins and arteries (known as vascular calcification), and bone mineral density in each image. “There are new things we’re discovering every day about the potential these images hold for disease prevention and detection,” says Dr. Dane, who oversees the technical component of harnessing CT scans for opportunistic imaging.
Dr. Bredella, who joined NYU Langone from Harvard Medical School in 2023, is using abdominal CT scans to measure the extent of vascular calcification. In a recent study, she and colleagues showed that the level of calcification measured in those scans accurately predicted a patient’s risk of cardiovascular disease, heart attack, and stroke.
Dr. Bredella is also using opportunistic imaging to quantify the amount of bone mineral density, muscle, superficial fat, and the deeper “bad fat” that surrounds abdominal organs, thereby revealing critical details about body composition that well-established measurements like body mass index, or BMI, can’t. “If you just measure someone’s BMI or waist circumference, you don’t know whether the fat is superficial or deep,” Dr. Bredella explains. “From extracted CT data, we can see not only how much fat there is, but also where it is.”
Another emphasis of NYU Langone’s radiological research is to accurately measure a patient’s muscle mass and the detection of sarcopenia, a condition in which muscle mass, strength, and performance are compromised due to advancing age or inactivity. Opportunistic quantification of muscle mass can forecast how well someone will fare after surgery or, say, a cancer diagnosis, findings that could eventually impact clinical decisions about treatment plans. “If we can use scans to quantify how much muscle patients have, how much they’re losing, and how quickly, that might influence the choice of certain drugs,” says Dr. Bredella.
Opportunistic imaging may even help address certain healthcare disparities. Medical education long taught that Black and Hispanic women were better protected against bone loss than White women, so they were less likely to be referred for bone mineral density tests. Using opportunistic imaging, Dr. Gyftopoulos can quantify a patient’s bone density and osteoporosis risk on CTs performed for routine clinical care.
In a study published in the September 2024 issue of the journal Bone, Dr. Bredella and other researchers showed that with the assistance of AI, CT imaging obtained for lung cancer screening can be used for osteoporosis screening. They also found that having more visceral fat, vascular calcification, and fatty liver disease were tied to bone loss. “Our study offers proof of concept that opportunistic screening could help with diagnosing osteoporosis in groups at greater risk, particularly patients who are elderly and those who smoke,” says Dr. Bredella.
Dr. Bredella believes opportunistic-imaging-based assessments could one day become an automatically generated add-on for routine scans, detecting signs of trouble that might otherwise go unnoticed. “I think we are close,” she says. “My goal and passion are to see it influence clinical decision-making and positively impact patient care."
