Autocorrelation and cepstral methods for measurement of tibial cortical thickness

Abstract

Cortical thickness of the tibia is related to stress fracture risk and overall skeletal status. Two methods are proposed for estimating tibial cortical thickness based on power spectra of ultrasonic echoes containing reflections from front and back surfaces of the cortex. The locations of the peaks in the autocorrelation function and the cepstrum are related to cortical thickness. Data were acquired on plastic plates in order to validate the methodology. These data indicate high correlations between estimated and true thickness with correlation coefficients r = 0.99, (95% confidence interval: 0.993-1.00) for the autocorrelation method and r = 0.99 (95% CI: 0.996-1.00) for the cepstral method. Data on six tibia samples in vitro indicate correlation coefficients of r = 0.92 (95% CI: 0.72-1.00) for the autocorrelation method and r = 0.85 (95% CI: 0.62-0.94) for the cepstral method. Estimates of precisions of the two methods were 0.3 +/- 0.1 mm (autocorrelation method) and 0.5 +/- 0.2 mm (cepstral method). One measurement in a human volunteer in vivo demonstrated clinical feasibility of the measurement and good agreement with cortical thickness assessed using peripheral quantitative computed tomography (QCT). This technology offers the promise of an inexpensive, fast, portable, simple, nonionizing technique for assessing skeletal status.