A deep learning algorithm can automatically detect lumbar vertebrae in MRI images, according to findings published in the Journal of Digital Imaging. This, the authors noted, has potential to improve clinician efficiency.
“Clinicians often diagnose lumbar disc herniation through physical signs in medical images, such as displacement and transformation,” wrote author Yujing Zhou, Nanjing University of Science and Technology in Nanjing, China, and colleagues. “A radiologist works redundantly to label every lumbar vertebra per patient for further diagnoses. Automatic vertebra detection can assist clinicians with etiological diagnoses, such as scoliosis, lumbar canal stenosis, and vertebra degeneration. This will help radiologists recognize each lumbar spine and relieve from duplicate works to annotate each new medical image.”
The authors’ algorithm uses a previous lumbar image to compare similarities between the vertebrae, a process they say is faster and saves memory. It was trained with the “massive” 2015 ImageNet Large Scale Visual Recognition Challenge dataset and tested on more than 2,000 cases, according to the study.
The algorithm had an accuracy of 98.6 percent and a precision of 98.9 percent. In addition, its specificity was 94.1 percent and sensitivity was 98.9 percent. While 2,701 MRI samples were read and labeled correctly by the algorithm, 38 samples featured errors.
“Most failed results are related to either a wrong S1 detection or a missed vertebra in MRI images,” the authors wrote.
Overall, however, the authors viewed their research as a success.
“The study demonstrates that a lumbar detection network supported by deep learning can be trained successfully without annotated MRI images,” they concluded. “It can be believed that our detection method will assist clinicians to raise working efficiency.”