Image Distortion in Biplanar Slot Scanning: Part 2 Technology-specific Factors

Biplanar digital slot scanning technology has become the standard of care in the treatment of scoliosis. Yet, the amount of distortion and reproducibility of this type of imaging modality has yet to be fully investigated. In our paper "Image distortion in biplanar slot scanning: part 1 patient-specific factors" we found that there was potentially clinically impactful interimage distortion. The purpose of this study was to evaluate the degree to which this image distortion was secondary to the image acquisition process.

Four 25 mm radio-opaque markers were placed at C3, T1, T12, and L5 on a full-length skeleton model. The skeleton was imaged in 10 different positions within the scanner. Five posteroanterior and 5 lateral images were obtained in each position. Two orthopaedic attending physicians and 3 orthopaedic resident physicians measured the markers for a total of 3200 measurements. Intraclass correlation coefficients (ICCs) and 95% confidence intervals were used to examine image distortion.

Average marker size was 24.77, with a standard error of measurement of 0.00493. Image distortion and standard error of measurement accounted for ∼0.5% to 1.5% of total the measurement. Overall, there was good reliability and consistency when looking at markers in different views (ICC 0.790), planes, and locations within the image. Horizontal measurements were found to be more consistent and have better reliability (ICC 0.881) than vertical measurements (ICC 0.386). Position within the scanner had minimal impact on the accuracy of the measurements.

This study demonstrates that there is minimal error due to image acquisition and measurement when using a biplanar slot scanner. Biplanar slot scanning technology tended to underestimate the size of the marker; however, the least accurate measurements only erred by 1.5% from the true length. This indicates that unlike traditional radiographs the sources of error in biplanar slot scanning images are not due to parallax and are likely due to patient-specific factors and rather than the technology itself.