Date of Graduation


Document Type

Dissertation (PhD)

Program Affiliation

Medical Physics

Degree Name

Doctor of Philosophy (PhD)

Advisor/Committee Chair

Stephen F. Kry, Ph.D.

Committee Member

David S. Followill, Ph.D.

Committee Member

Rebecca M. Howell, Ph.D.

Committee Member

Dragan Mirkovic, Ph.D.

Committee Member

Xinming Liu, Ph.D.

Committee Member

Francesco Stingo, Ph.D.


Many patients receiving external beam radiation therapy have metal implants that can affect their treatment, and these metal implants can degrade the accuracy of dose calculations. Dose calculation errors result from limitations of modern dose calculation algorithms in modeling metal/tissue interface effects. Metals also cause streak artifacts in the computed tomography (CT) images that are used for treatment planning, and these artifacts can also degrade dose calculation accuracy. Metal based-energy deposition kernels are a potential solution for the calculation errors associated with the limitations of the convolution/superposition (C/S) dose calculation method as they better model photon interactions and scatter in metals than water-based kernels, while CT metal artifact reduction methods have the potential to decrease calculation errors associated with imaging artifacts.

In this work, several metal-based energy deposition kernels (titanium, silver, and gold) were generated and characterized. These metal-based kernels exhibited more lateral scatter, more backscatter, and less energy deposited in the forward direction than water-based kernels, implying that simply scaling water kernels according to the local density encountered is inadequate for describing photon interactions in metals. These metal kernels were then implemented into a commercial collapsed cone C/S algorithm to investigate their dosimetric impact. In comparison to water-based kernels, metal kernels resulted in better prediction of the backscatter dose enhancement upstream of metals but decreased accuracy directly downstream of metals. When used for clinical dose calculations, the dosimetric benefit of using metal kernels was small (generally limited to a 0.5% decrease in calculation error for IMRT treatment plans).

In addition to metal kernels in the dose calculation algorithm, several commercial CT metal artifact reduction methods were investigated for their success in reducing dose calculation errors: the Philips O-MAR algorithm, GE’s monochromatic gemstone spectral imaging using dual-energy CT data (GSI), and dual-energy CT imaging with a dedicated artifact reduction algorithm (MARs). Each artifact reduction method was evaluated using several implants commonly encountered in radiation oncology (hip prosthesis, spinal fixation rods, and dental fillings), and its dosimetric impact was evaluated using two clinical cases. Though not always the most successful method, O-MAR was the most consistent and thus safest candidate for all-purpose metal artifact reduction in CT simulation imaging. GSI monochromatic imaging was beneficial for smaller, low Z implants but was not able to reduce the severe artifacts caused by larger, high Z implants and had very little effect on calculation accuracy. The MARs algorithm showed great success in certain scenarios (hip prosthesis and dental fillings) but also exhibited behaviors that are undesirable (i.e., metal distortion) and can actually result in increased errors in comparison to uncorrected CT images. Consequently, the MARs algorithm should be used with abundant caution for dose calculations.


photon therapy, convolution/superposition, energy deposition kernels, computed tomography, metal artifact reduction