header advert
Orthopaedic Proceedings Logo

Receive monthly Table of Contents alerts from Orthopaedic Proceedings

Comprehensive article alerts can be set up and managed through your account settings

View my account settings

Visit Orthopaedic Proceedings at:

Loading...

Loading...

Full Access

Spine

PRINCIPAL COMPONENT AND DISCRIMINANT ANALYSIS OF INTERVERTEBRAL MOTION PATTERN DIFFERENCES BETWEEN BACK PAIN SUFFERERS AND CONTROLS USING QUANTITATIVE VIDEOFLUOROSCOPY

The Society for Back Pain Research (SBPR), Northampton, England, November 2017



Abstract

Purpose and background

Identifying features in nonspecific low back pain (NSLBP) subjects that distinguish them from controls, or for elucidating subgroups, has proved elusive. Yet these would be helpful to monitor progress, improve management, and understand the nature of the condition. Previous work using quantitative videofluoroscopy (QF) has indicated that the distribution of motion between lumbar intervertebral joints is more uneven in those with a history of NSLBP. However, there maybe other features of these complex motion patterns yet to be revealed. A multivariate analysis was therefore carried out to explore other possible differences.

Methods and results

Intervertebral motion data of L2/3 to L4/5, from a previously published study was used. This examined 40 patients with NSLBP and 40 healthy controls, matched for gender, age and body mass index, who underwent passive recumbent QF in the coronal and sagittal planes. For each motion direction, principal components analysis was carried out and salient dimensions selected. Using a lower dimensional principal components (PC) representation, groups were compared using Hoteling's T test. Linear and quadratic discriminant analysis (LDA and QDA) was carried out using PC representations to examine group differences. The features most clearly distinguishing groups from the LDA was examined graphically. An analysis of the sensitivity of the results to the number of PC dimensions was carried out. The performance of the LDA and QDA classifiers were examined using leave-one-out cross-validation.

Conclusions

Hotelling tests revealed significant differences between groups for right and left side-bending. This was confirmed by LDA and QDA. There was no clear difference in the performance of these classifiers and performance did not improve by including more than 4 PC dimensions. Visualisation of the LDA indicated that patients had relatively lower amplitude motion at L4/5, compensated by higher amplitude at L2/3/4. These results point to additional features of lumbar motion that differentiates NSLBP.

No conflicts of interest

No funding obtained (however, the original study was funded by NIHR - CATCDRF09)


Email: