With more than 200 CT exams per 1,000 people performed every year in countries like France, the U.S. and Japan – and a total of 82 million in the U.S. alone – CT exams are an extremely common imaging modality. What is not common, however, is "opportunistic CT" – taking advantage of the secondary analysis of existing CTs to obtain quantitative body composition data without additional testing.
But with advances in automated measurements and normative values for bone and muscle health now published in large patient populations, this is beginning to change.
"Opportunistic CT presents an opportunity to add value to existing radiology exams by diagnosing common conditions that have a tremendous public health impact, like osteoporosis and sarcopenia," said Robert Boutin, MD, a radiologist at Stanford University School of Medicine.
Dr. Boutin noted that, according to the International Osteoporosis Foundation, after the age of 50, osteoporotic fractures occur in one out of three women and one of five men.
Although opportunistic CT has promise in improving diagnosis at no additional cost or radiation exposure to the patient, there have been concerns about scanner calibration and measurement interchangeability. Specifically, can similar measurements be obtained on CT scanners made by various manufacturers that are deployed throughout the U.S.?
To help address this gap in knowledge, radiologists and physicists at multiple institutions teamed up to investigate whether or not there is a systematic bias in reported CT numbers, as measured in Hounsfield Units (HU), when comparing the major CT manufacturers, using the American College of Radiology (ACR) phantom as the reference standard. The study analyzed more than 67,000 examinations acquired over seven years.
"CT number measurements between manufacturers have a systematic offset when compared to each other," Dr. Boutin said who discussed the study's results at a Monday session. "Knowledge of these offsets may be useful to harmonize HU values across platforms so as to optimize accuracy in the opportunistic diagnosis of osteoporosis."
Big Data, Big Opportunity
The study retrospectively collected ACR CT accreditation phantom data in a blinded fashion for four CT manufacturers. For each manufacturer, an adult abdomen CT technique was used to analyze the CT number for three materials: water, acrylic (surrogate for trabecular bone) and Teflon (surrogate for cortical bone). Comparisons were made to assess for systematic differences between CT manufacturers using a linear fixed effects regression model.
From this study, researchers noted that the CT number of water ranged from a mean of -0.3 to 2.7 HU, with mean differences between manufacturers that, although small, were highly statistically significant. Likewise, for the trabecular bone surrogate; mean differences in CT numbers across all manufacturers were small but significant. For the cortical bone surrogate, results showed highly significant mean differences in CT numbers across all manufacturers.
Poised for Paradigm Shift
These results show that CT number measurements compared between manufacturers have a systematic offset and these offsets change with the tissue being measured. "CT scanners made by different manufacturers show systematic HU offsets that are highly statistically significant," Dr. Boutin said. "The relatively small offsets relating to trabecular bone HU, however, supports the integrity of CT for the opportunistic diagnosis of osteoporosis."
According to Dr. Boutin, knowledge of these offsets may be used to harmonize HU values across platforms to optimize diagnostic precision. "The paradigm shift of using existing CT scans already obtained for other reasons to opportunistically screen for osteoporosis promises to substantially improve patient care," he added.