Multislice (MS-) CT has several advantages compared to single slice Spiral CT: It has a higher in plane spatial resolution, thinner sections may be obtained and these sections may be used to reconstruct volumes with any orientation in 3D space. The aim of this study was to use standard contrast-enhanced abdominal MS-CT datasets to generate BMD data and to compare these with BMD values obtained from standard quantitative CT (QCT).

50 patients undergoing contrast-enhanced abdominal MS-CT were recruited for this study (mean age: 50.5 ± 18.7 years). All patients were treated with chemotherapy for Hodgkin’s or Non-Hodgkin’s Lymphoma more than 12 months ago and underwent routine follow-up to exclude tumor recurrence. Since these patients were at higher risk for osteoporosis non-enhanced quantitative CT (QCT) was performed before the MS-CT examination. In the MS-CT datasets (parameters: collimation 16x0.75 mm, 200 mAs, 120 kVp) 1 mm sections (0.8 mm increment) were reconstructed in addition to the standard 5 mm sections. BMD was determined in volumes aligned parallel to the endplates of the vertebrae reconstructed from the 1 mm sections with different slice thicknesses (5 and 10 mm) and using different regions of interest (ROIs) (peeled and oval-shaped). BMD values obtained from MS-CT and QCT were correlated and a calibration equation was determined transforming MS-CT into QCT BMD.

The average BMD in the patients was 98.4 ± 45.7 mg/ml with QCT. Using contrast-enhanced MS-CT they were 134.4 ± 50.5 (oval ROI) resp. 132.1 ± 49.8 mg/ml (peeled ROI) for the 5 mm thick volumes and 134.5 ± 49.2 (oval ROI) resp. 131.8 ± 49.5 mg/ml (peeled ROI) for the 10 mm thick volumes. The corresponding coefficients of determination between QCT-BMD and MS-CT BMD values were r2= 0.89, 0.92, 0.92 and 0.93 (p<0.001). The conversion factor calculated for QCT-BMD and MS-CT BMD using the 10 mm thick volume with the peeled ROI was: QCT-BMD= 0.89 x MS-CT BMD – 20.9 mg/ml.

MS-CT BMD values may be transformed into QCT BMD data using a conversion factor based on a linear regression. Best results were obtained using reconstructed images with a slice thickness of 10 mm and a peeled ROI.