Purpose: Although several studies have reported on the application of biplanar stereo-radiographic technology in pediatric clinical practice, few have performed large-scale analyses. The manual extraction of these types of data is time-consuming, which often precludes physicians and scientists from effectively utilizing these valuable measurements. To fill the critical gap between clinical assessments and large-scale evidence-based research, we have addressed one of the primary hurdles in using data derived from these types of imaging modalities in pediatric clinical practice by developing an application to automatically transcribe and aggregate three-dimensional measurements in a manner that facilitates statistical analyses.
Methods: Mizzou 3D SPinE was developed using R software; the application, instructions, and process were beta tested with four separate testers. We compared 1309 manually compiled three-dimensional deformity measurements derived from thirty-five biplanar three-dimensional reconstructions (image sets) from ten pediatric patients to those derived from Mizzou 3D SPinE. We assessed the difference between manually entered values and extracted values using a Fisher's exact test.
Results: Mizzou 3D SPinE significantly reduced the duration of data entry (95.8%) while retaining 100% accuracy. Manually compiled data resulted in an error rate of 1.58%, however, the magnitude of errors ranged from 5.97 to 2681.82% significantly increased the transcription accuracy (p value < 0.0001) while also significantly reducing transcription time (0.33 vs. 8.08 min).
Conclusion: Mizzou 3D SPinE is an essential component in improving evidence-based patient care by allowing clinicians and scientists to quickly compile three-dimensional data at regular intervals in an automated, efficient manner without transcription errors.
Keywords: Biplanar radiography; Data extraction; Longitudinal assessments; Spine deformity.
© 2023. The Author(s), under exclusive licence to Scoliosis Research Society.