• Koji Tamai, Hidetomi Terai, Masatoshi Hoshino, Hitoshi Tabuchi, Minori Kato, Hiromitsu Toyoda, Akinobu Suzuki, Shinji Takahashi, Akito Yabu, Yuta Sawada, Masayoshi Iwamae, Makoto Oka, Kazunori Nakaniwa, Mitsuhiro Okada, and Hiroaki Nakamura.
• Department of Orthopedics, Osaka Metropolitan University Graduate School of Medicine, Osaka, Japan.
• Spine. 2023 Apr 15; 48 (8): 519525519-525.

Study DesignCross-sectional study.ObjectiveValidate the diagnostic accuracy of a deep-learning algorithm for cervical cord compression due to degenerative canal stenosis on radiography.Summary Of Background DataThe diagnosis of degenerative cervical myelopathy is often delayed, resulting in improper management. Screening tools for suspected degenerative cervical myelopathy would help identify patients who require detailed physical evaluation.Materials And MethodsData from 240 patients (120 with cervical stenosis on magnetic resonance imaging and 120 age and sex-matched controls) were randomly divided into training (n = 198) and test (n = 42) data sets. The deep-learning algorithm, designed to identify the suspected stenosis level on radiography, was constructed using a convolutional neural network model called EfficientNetB2, and radiography and magnetic resonance imaging data from the training data set. The accuracy and area under the curve of the receiver operating characteristic curve were calculated for the independent test data set. Finally, the number of correct diagnoses was compared between the algorithm and 10 physicians using the test cohort.ResultsThe diagnostic accuracy and area under the curve of the deep-learning algorithm were 0.81 and 0.81, respectively, in the independent test data set. The rate of correct responses in the test data set was significantly higher for the algorithm than for the physician's consensus (81.0% vs . 66.2%; P = 0.034). Furthermore, the accuracy of the algorithm was greater than that of each individual physician.ConclusionsWe developed a deep-learning algorithm capable of suggesting the presence of cervical spinal cord compression on cervical radiography and highlighting the suspected levels on radiographic imaging when cord compression is identified. The diagnostic accuracy of the algorithm was greater than that of spine physicians.Level Of EvidenceLevel IV.Copyright © 2023 Wolters Kluwer Health, Inc. All rights reserved.

24 keywords...

hide…