Structural health monitoring to detect the presence, location and magnitude of structural damage in cadaveric porcine spines

Structural health monitoring has been used successfully to identify defects in civil infrastructure and aerospace applications. Given that the majority of low back pain is thought to be mechanical in nature, our objective was to determine if structural health monitoring techniques could be employed successfully to identify the presence, location and magnitude of structural alterations within the spine. In six eviscerated cadaveric pigs, bone screws were drilled into the anterior bodies of L1-L5 and tri-axial accelerometers fixed to each spinous process. Vibration was then applied to the L3 spinous and frequency response functions obtained from each sensor axis before and after specific alterations of spinal structure. These alterations were produced at four unique locations and included (1) use of a cable tie to link anterior bone pins together and (2) progressive disc sectioning. Eighty percent of all data were used to train a neural network while the remaining data were used to test the network's ability to distinguish between structural states. The presence, location and magnitude of structural change within the spine was identified correctly in 50305040 possible neural network decisions. The diagnostic sensitivity and specificity of this technique ranged from 0.994 to 1.000. These results indicate that there is sufficient information embedded in frequency response data to identify the presence, location and magnitude of specific structural changes in the spine. If these techniques can be evolved for human use, structural health monitoring may provide a new approach toward understanding the underlying relations between spinal structure and function.