By Nick Klenske
When used in conjunction with other imaging, AI has the potential to provide radiologists with important information that may not be readily apparent from the scan alone.
“With the addition of AI tools, radiologists have a better chance of seeing the potential warning signs of an indolent disease that they may have otherwise missed,” said Angelo Scanio, BS, a medical student at UT Southwestern Medical Center in Dallas.
As a case in point, Scanio pointed to osteoporosis. This prevalent yet often undiagnosed condition, characterized by low bone mineral density (BMD), can lead to significant morbidity and mortality.
The gold standard for measuring BMD is dual-energy X-ray absorptiometry (DXA). “While osteoporosis is widespread, the challenge is that DXA is not consistently employed for all patients in need,” he explained.
Opportunistic screening, when supported by AI tools, could help fill this diagnostic gap.
During a Thursday presentation, Scanio discussed the results of a recent study on the opportunistic screening of osteoporosis using AI in chest CT examinations.
“The goal was to reduce the risk of severe osteoporotic fractures by integrating quantitative imaging data with supplementary clinical information,” he said.
Quantitative parameters from chest CT were extracted using Siemens’ AI Rad Companion tool. These were combined with clinical data and CT scan details in the Light Gradient Boosting Machine (LightGBM) model to predict low BMD, as defined by DXA scans.
The LightGBM machine learning model demonstrated robust performance in detecting low BMD, outperforming the sensitivity of radiological reports. More specifically, compared to the chest CT radiological reports, the model predicted low BMD with both a higher level of accuracy (79% vs 36%) and sensitivity (87% vs 12%). However, the AI model reported a lower specificity than the radiological reports (59% vs 94%).
According to Scanio, these results show that AI tools can help radiologists use CT scans that were obtained for unrelated reasons to detect an insidious disease early.
“The broader adoption of AI models in opportunistic screening for low BMD among patients undergoing chest CT examinations could become a life-saving method for detecting osteoporosis early – significantly reducing the morbidity, mortality, and health care related costs of those with the disease,” Scanio concluded.
Access the presentation, “Opportunistic Screening for Osteoporosis Using Artificial Intelligence in Chest CT Scans: Validation Against DEXA Scans,” (R1-SSCH09-4) on demand at RSNA.org/MeetingCentral.
© 2024 RSNA.
The RSNA 2024 Daily Bulletin is the official publication of the 110th Scientific Assembly and Annual Meeting of the Radiological Society of North America. Published online Sunday, December 1 — Friday, December 6.
The RSNA 2024 Daily Bulletin is owned and published by the Radiological Society of North America, Inc., 820 Jorie Blvd., Suite 200, Oak Brook, IL 60523.
