Can convolutional neural networks accurately select MRI protocols?
Deep learning-based convolutional neural networks (CNNs) can help radiologists select musculoskeletal MRI protocols, according to a study published by the Journal of Digital Imaging.
“The determination of the MRI protocol is an essential process in radiologic workflow,” wrote author Young Han Lee, MD, PhD, of Yonsei University College of Medicine in Seoul, South Korea. “The most appropriate protocol determination is essential for accurate radiologic interpretations and definitive radiologic decisions. However, this is time consuming and can result in a work burden for radiologists.”
The authors trained their CNN using more than 5,000 consecutive musculoskeletal MRI exams from a single institution’s electronic medical records (EMR). All exams were originally performed in 2016. Training took approximately 10 minutes.
To test the CNN, a dataset of more than 1,000 musculoskeletal MRI exams from January and February 2017 was taken from the same EMR. The CNN’s protocol choices were compared with what the patients’ radiologists had chosen.
Overall, the CNN’s accuracy was more than 94 percent. All pelvic bone, upper arm, wrist and lower leg MRI protocols were correctly determined.
“Deep-learning-based convolutional neural networks can be clinically utilized to determine musculoskeletal MRI protocols,” Lee concluded. “These results support using deep learning to assist radiologists in their work by providing timely and highly accurate protocol determinations that only require rapid confirmation. Furthermore, CNN-based text learning and applications could be extended to other radiologic task besides image interpretations, facilitating an improved work performance for radiologists.”
Michael has more than 18 years of experience as a professional writer and editor. He has written at length about cardiology, radiology, artificial intelligence and other key healthcare topics.