Image diagnosis of spinal disease using artificial intelligence

Abstract

Plain radiography of the spine is an essential method for the evaluation of spinal diseases. However, the accuracy of radiography for assessing spinal diseases is not optimal when image diagnosis of radiography for spinal diseases is highly specialized and often difficult for primary physicians. Skilled spine surgeons are not always available in a primary care setting, which is where most patients with spinal diseases are initially examined. Therefore, minimizing the diagnostic errors of spinal diseases by radiography is crucial. Image identification based on artificial intelligence (AI) has been recognized as being potentially effective as a diagnostic strategy for orthopedic diseases; however, the accuracy of detecting vertebral fractures (VF) and ossification of posterior longitudinal ligaments (OPLL) using AI has remained uncertain. In this study, the mean accuracies of AI for rhe detection of VF and OPLL were 86.0% and 98.9%, respectively. Thus, AI should be considered in the assistance of clinicians for the image diagnosis of spinal diseases.