AI-powered low-dose CT imaging is faster, safer, more accurate than other methods

A machine learning model produced low-dose CT images with greater speed and accuracy than previous attempts to use less radiation in CT imaging, according to a study published this week in Nature Machine Intelligence.

The model was developed by engineers from Rensselaer Polytechnic Institute in Troy, N.Y., and radiologists from Massachusetts General Hospital and Harvard Medical School, both in Boston. The deep learning algorithm sought to improve upon previous LDCT methods, which are safer for patients because of the reduced radiation, but often result in a corresponding reduction in image quality.

After being trained on the Mayo Clinic's LDCT Grand Challenge dataset and tested on chest and abdominal CT scans from Massachusetts General, the iterative reconstruction model produced images of equal or better quality than three popular LDCT imaging models, and did so significantly faster than the competing algorithms.

"This has radiologists in the loop," said Ge Wang, PhD, a professor of biomedical engineering at Rensselaer and a corresponding author on the paper. "In other words, this means that we can integrate machine intelligence and human intelligence together in the deep learning framework, facilitating clinical translation."

More articles about AI:
5 hospitals, health systems already implementing AI
IBM Watson Health, Medtronic app predicts hypoglycemic events with up to 90% accuracy
AI analysis of patient portraits helps diagnose rare diseases

© Copyright ASC COMMUNICATIONS 2020. Interested in LINKING to or REPRINTING this content? View our policies by clicking here.


Featured Webinars

Featured Whitepapers