Automated Detection of the Thoracic Ossification of the Posterior Longitudinal Ligament Using Deep Learning and Plain Radiographs

Figure 1

Preparation of images for training the object detection model. Images were annotated with a label [12] by manually inputting a minimal bounding box containing the OPLL on the thoracic lateral radiographs after the exact location of the OPLL was confirmed by CT to generate an image for the object detection training by one orthopedic spine surgeon (13 years): (a) dataset image of a plain lateral thoracic spine radiograph; (b) sagittal plane CT image to identify OPLL. CT: computed tomography; OPLL: ossification of the posterior longitudinal ligament.

(a)

(b)