Publication Date



Open access

Embargo Period


Degree Type


Degree Name

Doctor of Philosophy (PHD)


Biomedical Engineering (Engineering)

Date of Defense


First Committee Member

Ivaylo B. Mihaylov

Second Committee Member

Weizhao Zhao

Third Committee Member

Nesrin Dogan

Fourth Committee Member

Edward Dauer

Fifth Committee Member

Nelson Salas


Inverse radiotherapy optimization is based on a cost function that tries to minimize the radiation dose to volumes within a patient’s body. This dissertation explores the incorporation of electron/physical density information into the cost function. This can be termed dose-mass-based (DM) inverse optimization, as mass is the product of density and volume. Another approach for incorporating density in the optimization objective function is Energy-based optimization, where density is utilized to minimize energy deposition in mass (i.e. integral dose). The explorations herein included the investigation of sensitivity of mass-based inverse optimization with varying intensity modulation delivery parameters. The results of the study demonstrated that Energy optimization was significantly more sensitive than DM and dose-volume-based (DV) with respect to changes to both maximum segments per beam and minimum segment area. The second investigation considered the anatomical changes that occur to patients during radiotherapy. The dose-mass changes were compared between the planning CT and subsequent CTs, obtained mid-treatment and post-treatment. The results demonstrated that significant changes to dose-mass only occur for the target volumes and no statistically significant changes were observed for the surrounding normal anatomical structures. Another comparison was performed among plans developed with DV, DM, and Energy. The results showed that the anatomical changes yielded comparable differences regardless of the type of optimization used. Since density information is included in DM and Energy, plans the results suggest that the volumetric changes that occur dominate the density changes within the volumes. Under a third investigation, software tools were developed in order to calculate generalized equivalent uniform dose (gEUDs) and mass-weighted equivalent uniform dose (mgEUDs). mgEUD is mathematically more general than gEUD and in uniform-density media mgEUD transforms into gEUD. Incorporating physical density into the gEUD allows for a mass-weighted value representative of the uniform dose given to the mass rather than the volume. To further explore mgEUD patient outcome data for xerostomia of parotids and pneumonitis of lungs was used to correlate complication to mgEUD and in turn compare it to gEUD. The investigation determined that mgEUD values for the parotids did not show significant differences with respect to those of gEUD. In turn, lung mgEUD values demonstrated higher differences compared to values of gEUD. For radiation-induced pneumonitis of grade one and greater mgEUD showed lower standard deviations than those of gEUD. However, these differences did not translate into a better probability model of complication. The observed differences between gEUD and mgEUD using the Lyman-Kutcher-Burman normal tissue complication model were in the range of 2-3% for doses greater than 10 Gy. Incorporating density in inverse optimization plays a role in avoiding higher-density areas. This dissertation concluded that changes in inverse optimization delivery parameters indicated differences between volume-based and mass-based optimizations, but differences were not observed due to anatomical changes during radiotherapy treatment. The introduction of mgEUD demonstrated that there are differences in lungs with respect to gEUD and further investigation of normal tissue complication models may reveal a valuable correlation with treatment-related toxicity.


inverse optimization; density; gEUD; energy; mass; NTCP