The Determination of Lumbar Vertebral Fracture Characteristics Using Computed Tomography.

Abstract

In an effort to better underst and the causes of osteoporotic vertebral collapse, a computed tomographic method for measuring the regional distribution of bone mineral density in the lumbar vertebral centrum was developed. Fifty-nine in-vitro vertebrae (T12-L5) were removed within forty-eight hours of death from twenty-two cadavers. All underwent an initial regional bone mineral density measurement, were embedded in bone cement for mechanical test purposes, compressed to failure, and had the bone density measurement repeated. Each data set consisted of a "stack" of CT images which completely contained the vertebra of interest. A region-of-interest routine was applied to the initial mineral measurement data which numerically isolated eighteen separate cylindrical volumes in each centrum. The average density in each region was computed, and combined until a total of twenty-eight regional densities were calculated for each bone specimen. Multiple linear regression analyses on grouped data (by gender and cranial/caudal position) resulted in correlation coefficients from r = 0.424 to 0.915. Those individual regions which were most sensitive to ultimate load were located in the inferior and posterior areas of the centrum. Analyzing fracture patterns suggested that homogeneously low density in the superior and central thirds of the centrum may indicate a propensity to fail at significantly lower loads than other density patterns. The computed tomographic measurement method and numerical analysis techniques were also applied to 134 female volunteers between the ages of 40 and 60. Again a definite cancellous bone architecture was apparent, along with very non-uniform loss of bone. The areas which demonstrated the greatest rate of loss were those same regions which were identified as most sensitive to bone fracture load in the in vitro specimens.