The purpose of this study was to develop an algorithm that can be used to distinguish the central part of the vertebral body in an abdominal X-ray CT image, and to use that algorithm to automatically calculate three measures that are used to diagnose osteoporosis. In addition, the correlation between the BMD (bone mineral density) and these measures was investigated. We also examined whether the CT images obtained could be used to aid in diagnosing osteoporosis. The abdominal region contained in the third lumbar vertebra was scanned using spiral CT equipment (CT-W950SR: Hitachi). We judged whether female patients had osteoporosis using the diagnostic criteria available (Year 2000 revision, published by the Japanese Society for Bone and Mineral Research: only female data available). Twenty-two female patients, with an age range of 35-79 years (average age: 61.8), were examined. The mean CT number, coefficient of variation, and the first moment of the power spectrum were calculated as measures representing specific features of osteoporosis disease in the recognized vertebral body. We classified three measures obtained from the CT images for normal and abnormal groups using discriminant analysis, and the results obtained from the diagnosis criteria for the two groups were then compared. Our results showed that the algorithm could be used to distinguish the central part of the vertebral body in all cases, and to calculate these measures automatically. When the results of the discriminant analysis were applied to the three measures obtained from the CT images, the ratio usable for diagnosing a patient as osteoporotic (sensitivity) was 0.79 (11/14), and the ratio usable for diagnosing a patient as normal (specificity) was 0.63 (5/8). Therefore, in conclusion, we believe that this algorithm can be used to aid physicians in diagnosing osteoporosis, utilizing the measures obtained from abdominal X-ray CT images.
View full abstract