918 Hounsfield Units Shown in CBCT is Affected by Surrounding Tissue

Friday, March 23, 2012: 2 p.m. - 3:15 p.m.
Presentation Type: Poster Session
F. JIANG, Department of Mechanical Engineering, Indiana University Purdue University Indianapolis, Indianapolis, IN, and J. CHEN, Department of Mechanical Engineering, Indiana University Purdue University, Indianapolis, IN
Objective: Cone-Beam computed tomography (CBCT) provides bone image in terms of Hounsfield Units (HU) that may be used for estimating the bone mineral density (BMD). The objective was to test the errors from using the HU to predict BMD.

Method: Six CBCT scans from a specially designed phantom and five patient scans with a uniform BMD-rod attached to the patients’ faces were used. The phantom consists of simulated jaws with removable BMD-rods embedded in a background that simulates an average cross-section through an adult human head. Uniform BMD-rod should show uniform HU in theory. HU variations are the errors. The average HU values of different cross-sections along the long axis of the BMD-rods and the surrounding tissue were calculated. The surrounding tissue is defined within 80 mm from the BMD-rod’s center-line. Repeated measures regression was applied to test the relation between the HU of the BMD-rod and the surrounding tissue.

Result: The phantom scans showed an average of 708 HU with a standard deviation (STD) of 12 HU among the different sections of the uniform BMD-rod and of -428 HU with a STD of 26 HU for its surrounding tissue. The patients’ scans showed an average of 151 HU with a STD of 58 HU along the long axis of the BMD-rod and -672 HU with a STD of 67 HU STD for the surrounding tissue. The statistical analysis showed a strong inversed linear relation (R = -0.74) between the HU of the BMD-rod and surrounding tissue. The results demonstrate the error level of estimating the BMD from HU. However, relative BMD changes can be assessed as long as there are no significant changes in surrounding tissues.

Conclusion: The surrounding tissue affects the HU value. High HU variation in surrounding tissue results in a larger error in HU readings.

This abstract is based on research that was funded entirely or partially by an outside source: NIDCR R01-DE-018668

Keywords: CBCT and Digital image analysis