Assessment of osteoporotic vertebral fractures using specialized workflow software for 6-point morphometry

Purpose: To evaluate the time required, the accuracy and the precision of a model-based image analysis software tool for the diagnosis of osteoporotic fractures using a 6-point morphometry protocol. Materials and methods: Lateral dorsal and lumbar radiographs were performed on 92 elderly women (mean age 69.2 +/- 5.7 years). Institutional review board approval and patient informed consent were obtained for all subjects. The semi-automated and the manual correct annotations of 6-point placement were compared to calculate the time consumed and the accuracy of the software. Twenty test images were randomly selected and the data obtained by multiple perturbed initialisation points on the same image were compared to assess the precision of the system. Results: The time requirement data of the semi-automated system (420 +/- 67 s) were statistically different (p < 0.05) from that of manual placement (900 77 s). In the accuracy test, the mean reproducibility error for semi-automatic 6-point placement was 2.50 +/- 0.72% [95% CI] for the anterior-posterior reference and 2.16 +/- 10.5% [95% CI] for the superior-inferior reference. In the precision test the mean error resulted averaged over all vertebrae was 2.6 +/- 1.3% in terms of vertebral width. Conclusions: The technique is time effective, accurate and precise and can, therefore, be recommended in large epidemiological studies and pharmaceutical trials for reporting of osteoporotic vertebral fractures. (C) 2007 Elsevier Ireland Ltd. All rights reserved.