By: 16 September 2019
Utilisation of radiostereometric analysis (RSA) to assess quality of spinal fusion

Sara Parashin, Biomedical Engineer at the Orthopaedic Innovation Centre in Winnipeg, Canada, discusses a phantom study that assessed the feasibility of RSA as a technique to diagnose successful or failed spinal fusion

Spinal fusion is a surgical treatment to manage traumatic, degenerative, deformative and infectious diseases of the spine. Unfortunately, a portion of spinal fusions result in complications involving adjacent-segment degeneration and fusion failure. Medical device companies continue to develop technological advancements to improve implant performance and fusion rates. Despite these advances, the ability to diagnose fusion failure is a constant challenge. Accurate detection of fusion failure is especially important for health professionals to treat patients accordingly. Correctly identifying spinal fusion outcomes is also critical when assessing performance of new technologies and predicting long-term outcomes.

Currently, the gold standard for confirming fusion on cases of suspected failure is surgical exploration. While this method is highly accurate, it is invasive, timely, cost-prohibitive, and overall impractical. Non-invasive diagnosis measures have been explored, but also carry significant drawbacks such as poor radiographic accuracy and exposure to significant radiation.

Radiostereometric analysis (RSA) is a safe and highly accurate radiographic research technique that measures three-dimensional movement between two objects. RSA analysis requires the insertion of small (1.0 mm diameter) tantalum beads into the host bones being monitored for movement. RSA is known to have an accuracy much greater than conventional radiography, while using imparted radiation than plain X-rays. RSA began as a research tool in 1997 to detect skeletal growth and has progressed throughout the years, where the majority of studies measure implant stability in knee and hip joint replacements. 

The Orthopaedic Innovation Centre (OIC) is a research and engineering facility in Winnipeg, Manitoba, that dedicates 50 per cent of their work towards clinical RSA research studies. They have recently completed a phantom study that assessed the feasibility of RSA as a technique for diagnosing successful or failed spinal fusion.

The phantom study was developed as the first step towards creating a new, efficient and accurate way of diagnosing fusion to improve treatment plans.

The engineers at the OIC have been working with orthopaedic spine surgeons at the Health Sciences Centre Winnipeg in investigating the feasibility of RSA to assess cervical and lumbar spondylosis (fusion failure) by detecting micromovements between fused vertebrae (figure 1). Testing was performed using artificial (phantom) and cadaveric spine models. Three-level and two-level interbody fusions were performed by a surgeon on the cervical and lumbar regions, respectively. Several RSA beads (4-8 per vertebra) were implanted during the fusion process that maximised bead spread within the bone (Figure 2 a and b).

Feasibility of RSA was assessed by measuring bead placements, precision and accuracy. RSA bead positioning is a critical influencing factor for RSA accuracy and precision. Therefore, an optimal bead placement plan for cervical and lumbar regions was developed. Bead placements were determined by bead distribution values and bead visibility on the RSA images. Precision of RSA in spine was measured by taking double RSA examinations in sequential order with slight model displacement between exams. Precision was calculated as the standard deviation of migration between double exams. Accuracy of the system was measured in longitudinal and sagittal translation, and transverse rotation, and calculated as the mean difference between known and measured values.

RSA exams were taken using dual ceiling mounted X-ray sources that intersect at the area of interest and capture a static image pair. RSA analysis was carried out using a model-based software that re-generates the joint segments in three-dimensional space.

Good bead visibility and detectability was found among the RSA images in model-based software (figure 3 a and b). Between cervical and lumbar fusions, bead detectability was slightly better in cervical segments rather than in lumbar segments. The reason for minor bead loss in the lumbar region is attributed to implant and bead overlap on RSA images. Fusion hardware is larger in lumbar than cervical, creating greater obstructions.

Although bead detectability was greater in cervical fusion than lumbar, bead spread showed the opposite effect. Achieving adequate bead spread among cervical vertebrae was more challenging than in lumbar vertebrae simply due to their small size. Smaller areas reduce possible locations for beads and decrease bead distribution by forcing bead collinearity, resulting in reduced precision and accuracy.

Longitudinal and sagittal translational precision was found to be <0.82 mm in cervical segments, and <0.25 mm in lumbar segments. While transverse rotational precision was found to be <1.89° in cervical segments, and <0.40° in lumbar segments.

Precision results were better in the phantom models than in the cadaveric models. This can be explained by tissue attenuation presented in cadaveric models and reduced bone quality. Despite these differences, precision across all models proved to be adequate for measuring micromovement between fused vertebrae in cervical and lumbar fusion procedures.

Accuracy was measured in the phantom model only. Higher accuracy was found in the lumbar segments (<0.02 mm) than cervical segments (<0.10 mm) for sagittal and longitudinal translations. Transverse rotational accuracy was higher in cervical (<0.10°) than lumbar (<0.22°).

Overall, the results from this work indicate that RSA is a feasible method to detect micromotion in cervical and lumbar regions following spinal fusion in artificial and cadaveric spine. There are however, limiting factors to this research method.

A posterior surgical approach was used in the present phantom study, therefore bead positioning was limited to posterior aspects of the spine. Assessment of precision and accuracy though artificial and cadaveric modelling is a good starting point for research but fails to consider factors present in a clinical environment, such as effect from tissue attenuation, clinical scenario complications, and reduced bone quality from disease. As a result, it is presumed that RSA precision and accuracy will likely be reduced clinically, but remain quite feasible.

For these reasons, further research is needed to validate RSA precision and accuracy in the spine. Fortunately, the findings presented put RSA in good standing to address a significant void of gold-standard diagnostics in detecting whether spinal fusions are successful or have failed. The clinical measurement would be to classify a threshold of intervertebral movement as non-union (fusion failure) and with the converse being a union (successful fusion). Accurate diagnosis of fusion will reduce costs, prevent unnecessary revised surgical intervention, and most importantly improve patient treatment.

This research supports RSA as a research tool to a similar affect, by assessing spine biomechanics following non-instrumented decompression surgery. This application would measure spinal motion and correlate the findings to clinical outcomes to help determine the true cause of recurring back pain following spine surgery. Where the reason for recurring pain may be abnormal spinal movement, psychologic, or chronic.


Sara Parashin is a biomedical engineer at the Orthopaedic Innovation Centre (Winnipeg, Manitoba) who dedicates her time to orthopaedic research studies, engineering testing and explant retrieval analysis. Sara began working for the company in 2015 and now leads the biomechanical testing and clinical research departments related to the spine.