Reliability and accuracy of ultrasound measurements with and without the aid of previous radiographs in adolescent idiopathic scoliosis (AIS)

The objectives of this preliminary study were to assess the reliability and accuracy of ultrasound (US) for measuring coronal curvature with and without the aid of a previous radiograph, and to evaluate the ability of US to detect curve progression in adolescent idiopathic scoliosis (AIS) patients. Four raters measured 20 AIS US images twice at one-week intervals. Intra-rater reliability and correlation with radiograph were investigated with (rater 1) and without (raters 2-4) the aid of a previous radiograph. The center of lamina (COL) method was used to approximate the Cobb angle. Thirty-six curves were identified. All raters showed high intra-rater reliability (ICC[2,1] > 0.80). With the aid of a previous radiograph, rater 1 showed higher correlation with radiograph (ICC[2,1] = 0.86), better standard error of measurement (SEM = 2.2A degrees), and improved error index of selecting end-vertebrae (EI = 1.34), but no statistical improvement of intra-rater reliability (p > 0.05). For rater 2-4, the range of the ICC[2,1] values between US and radiograph measurements, the SEM value, and the range of the EI values were 0.70A degrees-0.72A degrees, 3.3A degrees, and 1.65A degrees-2.36A degrees, respectively. Specificity and sensitivity of US for detecting curve progression were 0.91 and 0.83, respectively. Using a previous radiograph as a measurement aid helped the user to measure coronal curvature from US images, and improved the accuracy of end-vertebrae selection. US showed high sensitivity and specificity for detecting curve progression, indicating that US may be a suitable, radiation-free alternative for monitoring patients with AIS who have mild or moderate curves.