Abstract
Introduction
Accurate radiographic measurement of sagittal alignment is essential for evaluating adult spinal deformity (ASD) and pre-operative planning. However, the limited capabilities of traditional picture archiving and communication systems (PACS) often necessitate rudimentary techniques and estimations of anatomic landmarks and angles. Though dedicated spine measurement software (SMS) has been studied and validated, there are no direct comparisons PACS to SMS.
Material and Methods
Eleven independent observers (7 surgeons, 4 researchers) with varying levels of experience digitally measured 20 primary and revision ASD patient radiographs for pelvic incidence (PI), pelvic tilt (PT), lumbar lordosis (LL), PI-LL mismatch, thoracic kyphosis (TK), and sagittal vertical axis (SVA) in 2 rounds. Round 1 used basic line and angle tools in traditional PACS; Round 2 used the sagittal alignment tool in a validated software dedicated to spine measurement and operative planning. SMS automatically calculates spino-pelvic parameters from 6 user-identified anatomic landmarks, including outlines of femoral heads and vertebral endplates. Results were analyzed for means, standard deviations, coefficient of variation (CV), and intra-class correlation (ICC).
Results
Mean values for PACS measurements were significantly greater than SMS for PI, PT, PI-LL, and TK (all P < 0.010), though differences were within previously described margins of error. The standard deviations were also significantly larger for the PACS measurements in all parameters (P < 0.012 for all) except TK. Excluding TK, the variation in measurement was significantly greater for PACS (CV=14–34%) versus SMS (CV=11–23%). The ICC values for all parameters were greater than 0.64, and when PI was excluded, all were greater than 0.92. Inter-rater reliability was greater in SMS compared with PACS for nearly all measurements: PI, PT, PI-LL, LL, and SVA. For both SMS and PACS, the lowest ICC was observed in PI, and the highest ICC was seen in SVA. The parameters with the greatest differences in inter-rater reliability between PACS and SMS were PI (PACS ICC: 0.647 vs SMS ICC: 0.810) and PI-LL (PACS ICC: 0.921 vs SMS ICC: 0.970). TK had the most similar ICC values between PACS (0.955) and SMS (0.945), and was the only parameter for which the PACS ICC was greater than the SMS ICC. When only the surgeons' measurements were considered, the differences between PACS and SMS ICC were substantially greater. Among the surgeons, SMS had higher ICC than PACS for all parameters (ex. PI-LL: 0.957 vs 0.896). PI still had the lowest inter-rater reliability (PACS ICC: 0.505 vs SMS ICC: 0.752) and SVA had the highest (PACS ICC: 0.985 vs SMS ICC: 0.994).
Conclusion
SMS measurements provide significantly more accurate and reliable measurements with less variation than PACS. The greater reliability of SMS is amplified in surgeon-only analyses, demonstrating the clinical utility of SMS versus traditional PACS. Accurate interpretations of sagittal alignment are critical because poor measurements may lead to insufficient or overly aggressive operative plans, and thus undesirable clinical results. Consistent use of SMS in the clinical evaluation and operative planning of ASD patients would be advantageous given the significant differences in values, variance, and reliability between PACS and SMS.