Automated measurement of spinopelvic alignment parameters using a spine planning software: a validation study

TL;DR

This study validates a software tool for measuring spinal alignment, finding it accurate and faster than manual methods but with some limitations.

Contribution

The study validates automated spine planning software for measuring spinopelvic alignment with high reliability and speed.

Findings

Automated measurements showed high reliability (ICC = 1) and strong agreement with manual methods.

Automated measurements were significantly faster (62 s vs. 227 s) but failed to measure certain parameters like SVA in many cases.

Manual measurements had reliability ranging from fair to excellent (ICC 0.44–0.99).

Abstract

Accurate assessment of spinopelvic alignment is essential for managing adult spinal deformities. This study validates the Brainlab Elements Spine Planning software for automated measurement of spinopelvic parameters, comparing it with manual methods.

Spinopelvic parameters were measured manually and with the software in 21 patients with degenerative spinal disease, including instrumented and non-instrumented spines. Accuracy, intraobserver, and interobserver reliability were evaluated using Bland–Altman plots and intraclass correlation coefficients (ICCs). Measurement times were also compared.

The software showed high reliability (ICC = 1), while manual measurements ranged from fair to excellent reliability (ICC 0.44–0.99). Bland–Altman plots indicated strong agreement between automated and manual measurements, though variability was noted for certain parameters. Automated measurements were significantly faster, averaging 62 s versus 227 s in manual measurements (p < 0.001), though 76.2% of cases had at least one parameter that could not be measured automatically, most frequently the sagittal vertical axis (SVA) and several coronal parameters.

The Brainlab Elements software provides accurate, reproducible, and time-efficient spinopelvic measurements for parameters it successfully captures. However, frequent failures in assessing SVA and coronal plane parameters automatically suggest that further refinement of the software is necessary.

Figures