Lumbar Vertebrae Morphological Analysis and an Additional Approach for Vertebrae Identification in Lumbar Spine DXA Images.

INTRODUCTION: Bone mineral density measured by dual-energy X-ray absorptiometry (DXA) is the gold standard for diagnosis of osteoporosis. However, DXA interpretation can be influenced by analytic errors which can, in turn, interfere with diagnostic accuracy. One such source of potential error is the accurate identification of specific lumbar vertebrae on lumbar spine DXA images. Although the International Society for Clinical Densitometry has introduced several approaches to aid with this process, there remain individuals whose lumbar vertebrae cannot be labeled with certainty using these approaches. We designed the present study to systematically evaluate lumbar spine DXA images among a large cohort of Chinese patients and present an additional simple strategy for identifying L5 named the "em dash"-shaped L5 or "-"-shaped L5, based upon the two-dimensional relationship between the bilateral pedicles with the central spinous process on DXA images.

METHODOLOGY: Lumbar spine posteroanterior DXA images from adult patients receiving care at a large tertiary hospital in Beijing, China from May to August 2016 were retrospectively reviewed. For each patient, data were collected regarding key anatomic features seen on DXA (positions of the most superior portions of the iliac crests, the lowest vertebra with ribs, and the longest transverse processes) and the proportion of patients presenting with "H"-shaped L4 and "-"-shaped L5. Chi-squared analyses were used to compare proportions across age strata.

RESULTS: DXA images from 1125 patients (79.6% female) were evaluated. The mean age of patients was 52.5 ± 14.8 yr (range: 19-90 yr). A horizontal line drawn across the superior-most portions of the iliac crests crossed the disk space between L4 and L5 among 78.3% of patients. The lowest ribs were most frequently (83.9%) observed at T12. Almost 80% of individuals had the longest transverse processes at L3. L4 was predominantly "H"-shaped (73.3%), however we found that the proportion of individuals with "H"-shaped L4 decreased steadily after 50 yr of age (p < 0.001). By contrast we observed that L5 was predominantly "-"-shaped (80.3%), with no significant differences in proportions across all age strata (p = 0.063).

CONCLUSIONS: The "-"-shaped L5 can be incorporated as an additional reference tool for lumbar spine DXA image analysis and may be particularly helpful for lumbar vertebrae identification among patients over 50 yr of age.