A PRACTICAL EVALUATION OF 5-DOSE-VOLUME HISTOGRAM REDUCTION ALGORITHMS

A unifying approach to cumulative dose-volume histogram (CDVH) reduction analysis is presented, utilising two weighted linear interpolation models (VWD, DWV), two weighted probability models (VWP, DWP) and a novel integral probability model (IPM). As a test of their predictive value these algorithms were applied to CDVH data generated from lung doses, measured by TLD arrays in a female anthropomorphic phantom. Three arbitrary configurations of breast size and location of "target volume" within the breast were "treated", using an appropriate electron field (Varian Clinac 1800) or double-plane iridium-192 implant. Calculated effective doses from each of the reduction algorithms showed the iridium implant to be dosimetrically the most favourable in two the three configurations. Likewise, complication probabilities, based on a logistic dose-volume response function showed lung complication probabilities to be lower for the interstitial technique in the same situations. All algorithms tested showed reasonable consistency, with the exception of the VWD. The rationale and value of comparative rather than absolute dose-volume histogram analyses are discussed.