Publication Date


Document Type


Committee Members

Thomas N. Hangartner (Advisor), David F. Short (Committee Member), Julie A. Skipper (Committee Member)

Degree Name

Master of Science in Engineering (MSEgr)


The measure of bone strength and fracture risk assessment is based primarily on bone mineral density (BMD). Computed tomography (CT), a frequently used radiographic technique that provides cross-sectional images, can be used for BMD quantification. Called quantitative computed tomography (QCT), it determines true volumetric density and also differentiates between trabecular and cortical bone. As a result, QCT improves the ability to detect early changes in trabecular BMD accurately and precisely.

The introduction of helical CT for quantitative analysis introduces new complexities. Even though using multi-detector helical CT for BMD assessment is being suggested, a detailed analysis of its validity is missing from the literature. The current project exhaustively considers the various aspects of the problem. First, a detailed overview of the problems incurred while using multi-detector helical scanning for QCT is given and a suite of tests to evaluate MSCT machines is introduced. For this purpose a rod phantom was designed to mimic a patient, with various amounts of bone, and scans of the rod phantom along with the Mindways calibration phantom were taken using multi-detector spiral CT scanners from the three major CT scanner manufacturers General Electric (GE), Siemens and Toshiba. 4-, 16- and 64-slice scanners were used for the study.

Except for the 16/64 slice Siemens scanners, an underlying cyclic pattern was observed in the data collected using the remaining 4 scanners in this study. Depending on the interpolation algorithm, the frequency of this pattern was either equal to or half of the total number of detector arrays. The numbers for precision, in mg/cc, returned from this experiment were within the range of 0.0053 - 7.8121 mg/cc. Variations of the density with pitch were small and did not extend the precision range specified by the previous experiments.

Spanning the entire allowable scan table length, a shift ranging from 0.0022 to 9.3 HU and 0.0013 to 15.02 mg/cc was observed for the GE and Siemens CT scanners when transitioning to the table extension. For the Toshiba scanner, the shifts in the range of 61 to 479 HU were reduced to 0.63 to 9.73 mg/cc after calibration. The table-height experiment revealed that the CT numbers and hence the calibrated densities are a function of the position of the object being scanned within the gantry. The change in the densities with table height is reduced after calibration, with the low-density ROIs being an exception.

An averaging calibration method to take into account the cyclic repetition appears to be reasonable to allow for reduced errors using spiral CT for quantitative bone analysis. Even though the calibration method compensates for the variations because of table position and height, scanning close to the isocenter within the gantry and avoiding inclusion of the table extension in the scan field of view will be beneficial. Spiral CT can be used for quantitative purposes by carefully setting the scan protocol and applying proper corrections.

Page Count


Department or Program

Department of Biomedical, Industrial & Human Factors Engineering

Year Degree Awarded