The vibration and acoustic emissions produced within facet joints of the lumbar spine, known as crepitus, can be a potential biomarker to identify decreased joint functioning and the site of low back pain. Past analyses of these data with human observers had been excessively time consuming for any eventual practical clinical application, and a more expedient algorithmic method of analysis was sought. Using piezoelectric accelerometers and a silicone mechanical “phantom model” simulating an in vivo human lumbosacral spine, we attempted to identify the site of crepitus. Ten accelerometers were fitted to the phantom model, with six indicating spinal level (L1-S1) and four indicating laterality (right and left, at the L1-L2 and L3-L4 levels). An impulse was provided to the phantom model at each of the ten spinal joint sites of varying level and laterality, using a mechanical impulse delivery device. To assess reliability and validity, each joint was struck with an impulse three separate times. Each impulse was transduced by ten piezoelectric accelerometers and recorded by LabView in a TDMS file format. Three strikes on each joint site produced a total of thirty recordings, which were first randomized. Recordings were then analyzed &nbsp;by two human observers and the automated method program. The experiment was replicated (with re-randomization) one week later. Human observers and the operator using the automated method were blinded to each other and their previous work. The automated method was programmed using MATLAB. Aberrant noise was filtered by means of the Curve Fitting and Signal Processing toolboxes, using a high-pass filter of uniform cutoff frequency of 8 kHz across all signals. The location of crepitus in terms of both the level and the side (left or right) was determined by establishing a threshold based on moving one percent outside the baseline signal extrema for each recording, which we termed “dynamic threshold”. At least six of ten accelerometers passing outside the dynamic threshold at a time, for a recording, indicated the presence of crepitus. The first of the laterality accelerometers which passed outside its dynamic threshold indicated the side, and the first two accelerometers indicating the level that passed outside the threshold indicated the segment level of the crepitus event. In cases where two accelerometers indicating level were not adjacent to one another, the program obtained the third level accelerometer and determined which of the first two accelerometers it was adjacent to. The two accelerometers adjacent to one another, thereby, indicated the level. Both the intra- and inter-rater reliability and validity (accuracy), using a weighted Kappa statistic were determined. In addition, a comparison of the time to completion for each of the human observers and automated method was made using ANOVA. Results from Run 1 validity, intra- and inter-rater reliability were all 0.983 (0.968-1.0) or greater, with any weighted Kappa statistic above 0.900 indicating almost perfect agreement. The average time to completion for a run was 3,810.00 seconds (+296.98) for the first human observer; 2,940 seconds (+763.68) for the second observer; and 81.50 seconds (+21.92) for the automated method. The average time to completion for a run of the automated method was found to be significantly less than the average time to completion for a run of both human observers (ANOVA p=0.009). Future work is planned to evaluate the reliability and validity of the automated method using prospective data from human subjects.

Vibration and Acoustic Crepitus Sensing Using Piezoelectric Accelerometers and Automated Signal Analysis

Paper Proceedings: https://asmedigitalcollection.asme.org/IMECE/IMECE2021/volume/85543

ASME 2021 International Mechanical Engineering Congress and Exposition (IMECE2021)

Supporting the need among researchers worldwide for additional online and multimedia content, ASME now offers the ASME Conference Video Collection. This platform will collect and provide long-term access to conference presentation videos, encompassing all technology areas within ASME's technical conference and events program. About 30,000-40,000 engineers across all mechanical engineering and related disciplines participate in ASME conferences each year. This includes authors of approximately 6,000 to 8,000 technical papers published by ASME, the majority of whom also submit video presentations for our virtual conferences. The video collection will be available as an add-on to existing product packages or as a separate subscription.