Researchers use deep learning to detect cartilage lesions in knee MR images
Deep learning technology can be used to evaluate MR images of the knee, according to a new study published in Radiology.
“This study was performed to determine the feasibility of using the deep learning approach to detect cartilage lesions within the knee joint on MR images,” wrote lead author Fang Liu, PhD, with the department of radiology at the University of Wisconsin School of Medicine and Public Health in Madison, Wisconsin, and colleagues. “It was our hypothesis that the deep learning method would provide similar diagnostic performance to and higher intraobserver agreement than the interobserver agreement of clinical radiologists for detecting cartilage softening, fibrillation, fissuring, focal defects and diffuse thinning due to cartilage degeneration and acute cartilage injury.”
The authors developed a fully automated deep learning-based cartilage lesion detection system using two deep convolutional neural networks (CNNs). The first CNN focused on segmentation of cartilage and bone in the images, and the second CNN evaluated structural abnormalities detected within the cartilage. Two individual evaluations of the system were performed for the study.
Overall, the detection system had a sensitivity of 84.1 percent and a specificity of 85.2 percent for evaluation 1. For evaluation 2, the sensitivity was 80.5 percent and specificity was 87.9 percent. Areas under the receiver operating curve were 0.917 for evaluation 1 and 0.914 for evaluation 2.
“The cartilage lesion detection system was able to detect all types of cartilage lesions, including softening, fibrillation, fissuring, focal defects, and diffuse thinning,” the authors wrote.
Liu et al. also compared these numbers to the diagnostic performance of radiology residents, fellows and a fellowship-trained radiologist with 17 years of clinical experience. They had a sensitivity that ranged from 60.8 percent to 80.2 percent and a specificity that ranged from 92.2 percent to 96.5 percent.
“The sensitivity and specificity of the cartilage lesion detection system were comparable to the diagnostic performance of clinical radiologists, including radiology residents, musculoskeletal radiology fellows and a musculoskeletal radiologist,” the authors wrote. “The high AUCs of the cartilage lesion detection system were similar to the AUCs for deep learning techniques used in previous studies to detect pulmonary nodules at chest CT, classify pulmonary tuberculosis at chest radiography and breast density at mammography, detect coronary artery stenosis at contrast material-enhanced chest CT and prostate cancer at pelvic MRI, and grade the severity of osteoarthritis at hip radiography, which further emphasizes the promising preliminary results of computer-based methods for evaluating medical images.”
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.