Artificial intelligence (AI) may be able to reduce the amount of gadolinium patients are exposed to during MRI scans, according to research presented Monday, Nov. 26, at RSNA 2018 in Chicago.
“There is concrete evidence that gadolinium deposits in the brain and body,” lead author Enhao Gong, PhD, a researcher at Stanford University in Stanford, California, said in a prepared statement from RSNA. “While the implications of this are unclear, mitigating potential patient risks while maximizing the clinical value of the MRI exams is imperative.”
Gong and colleagues trained a deep learning algorithm with MRI images from 200 patients who had undergone contrast-enhanced MRI exams. Three sets of images were gathered from each patient: pre-contrast scans, low-dose scans acquired once 10 percent of the standard dose had been administered and full-dose scans acquired after 100 percent of the standard dose had been administered. The algorithm could then use the zero-dose and low-dose images to “approximate” the full-dose images, saving the patient from receiving that dose.
The researchers noted the image quality of the AI-enhanced MR images was “not significantly different” from the full-dose scans that actually represented a full dose of gadolinium.
“Low-dose gadolinium images yield significant untapped clinically useful information that is accessible now by using deep learning and AI,” Gong said in the same statement.
The team’s next step will be to study this concept further, testing it with a variety of contrast agents and “across a broader range of MRI scanners.”