RSNA 2017 Artificial Intelligence Preview

Deep learning can predict stenosis on fast SPECT-MPI

By Erik L. Ridley, staff writer

November 6, 2017 --

Friday, December 1 | 11:00 a.m.-11:10 a.m. | SST02-04 | Room E450A
Researchers have found that deep learning can improve the detection of potentially significant ischemic defects on raw, high-speed SPECT myocardial perfusion imaging (MPI) studies.

SPECT-MPI is widely used in the U.S. for diagnosing coronary artery disease (CAD); more than 8 million scans are performed nationwide each year, according to presenter Julian Betancur, PhD, of Cedars-Sinai in Los Angeles. In a multicenter study, Cedars-Sinai researchers evaluated the use of deep convolutional neural networks (CNNs) and quantitative polar maps for automatically predicting the presence of obstructive coronary stenosis from raw, fast MPI data.

"Our study is the first to introduce a deep-learning application to MPI and demonstrates an enhancement in the predictive value of fast SPECT-MPI," Betancur told "A fully developed and tested tool like the one proposed can be deployed as part of SPECT-MPI display and interpretation packages, with important implications in clinical routine."

How did they accomplish this level of performance? Attend this presentation on Friday morning and you'll get all the details.