Improvement on the Accuracy and Reliability of Ultrasound Coronal Curvature Measurement on Adolescent Idiopathic Scoliosis With the Aid of Previous Radiographs.

Abstract

Retrospective study of the coronal curvature measurement on ultrasound (US) images with the aid of previous radiographs. To compare the reliability and accuracy of the coronal curvature measurements from US images on children who have adolescent idiopathic scoliosis (AIS) with and without the knowledge of previous radiographs. Using US imaging technique to measure coronal curvature on children with AIS has demonstrated high intra- and interrater reliabilities. However, the selection of end-vertebrae and the measurement difference between radiography and the US method were only moderately reliable. Two raters measured the coronal curvatures from 65 AIS standing US spine images, without (measured one time) and with the aid of previous standing radiographs (measured two times). The intra- and interrater reliability, the correlation and the difference between the radiographic and US measurements, and the error index of the end-vertebrae selection were assessed. Overall, 109 curves were investigated. The intraclass correlation coefficients (ICC) of intra- and interrater reliability of the US coronal curvature measurement with the aid of previous radiographs (AOR) were 0.95 and 0.91, respectively. In comparison with the radiographic measurements, the correlation of AOR method (R) was 0.90 and the MAD was 2.8°; the corresponding results of the US measurement without the AOR (blinded US method) were 0.73° and 4.8°, respectively. The average error index on end-vertebral selection improved 43% with the AOR. The AOR method significantly improved reliability and accuracy of the spinal curvature measurement on US images compared with the blinded US method (P<0.001). It indicates that US standing images with the AOR can be used as a reliable and accurate nonionizing imaging method to monitor children with AIS. 3.