番茄社区

CHICAGO – Low-dose gadolinium coupled with a deep learning algorithm produced magnetic resonance images that were non-inferior to a full dose of the contrast agent, researchers reported here.

Qualitative ratings showed non-significant differences between the proposed method – a 10% level of gadolinium dose – and acquired full-dose contrast-enhanced MRI images, which was also verified with non-inferiority testing, said Enhao Gong, PhD, a founder of Subtle Medical of Menlo Park, California, a medical imaging software company that uses artificial intelligence to enhance medical imaging quality.

In Gong's presentation at the , he said the success of finding images that were non-inferior to those of full-dose gadolinium has spurred the researchers to push the algorithm further – perhaps to 5% or maybe even down to 0.

"There is concrete evidence that gadolinium deposits in the brain and body," he told 番茄社区. "While the implications of this are unclear, mitigating potential patient risks while maximizing the clinical value of the MRI exams is imperative."

While working at Stanford University in California, Gong and colleagues developed artificial intelligence techniques that teach computers by examples. Through use of models called convolutional neural networks, the computer can not only recognize images but also find subtle distinctions among the imaging data that a human observer might not be capable of discerning.

For the study, the researchers used magnetic resonance images from 200 patients who had received contrast-enhanced MRI exams for a variety of indications. The team collected three sets of images for each patient: pre-contrast scans, done prior to contrast administration and referred to as the zero-dose scans; low-dose scans, acquired after 10% of the standard gadolinium dose administration; and full-dose scans, acquired after 100% dose administration. Each patient had all three scans taken, and those images were then submitted to experienced readers for evaluation.

Gong said the algorithm learned to approximate the full-dose scans from the low-dose images, and neuroradiologists then evaluated the images for contrast enhancement and overall quality.

"The low-dose images were very similar to the high-dose images," said one of the readers and a co-author of the study, Max Wintermark, MD, chief of neuroradiology at Stanford. Wintermark told 番茄社区 that he could discern which of the scans was low-dose, "but the information extracted was pretty much the same. I wasn't getting less information from the low-dose scans.

"They did all kinds of MRI scans, and I reviewed the results," he said. "These results still need to be validated."

Asked for her perspective, Cindy Lee, MD, of New York University Langone Health, who was not involved with the research, said: "Gadolinium goes in, but it doesn't come out. We don't know if gadolinium deposits cause any problems or less of it – or none of it would be beneficial, we think. Right now in clinical practice we can reduce the dose by 50% if the patient has liver or kidney disease. There definitely is the potential for going lower to decrease the chances of getting deposits in the brain or the rest of the body."

Lee added that she would like to see the images produced with the lower levels of gadolinium contrast and applauded the idea that even lower doses are being contemplated: "Zero, though, is going to be hard to achieve," she said.

Gong explained that low-dose gadolinium images yield "significant untapped clinically useful information" that is accessible by using deep learning and artificial intelligence. Now that he and his colleagues have shown that the method is technically possible, they want to study it further in the clinical setting, where Gong said he believes it will ultimately find a home.

Future research will include evaluation of the algorithm across a broader range of MRI scanners and with different types of contrast agents. "We're not trying to replace existing imaging technology," he said. "We're trying to improve it and generate more value from the existing information while looking out for the safety of our patients."