Using 54 patient plans, the relationship between the treatment planning system (TPS) TRAK (TRAKTPS) and the estimate and planimetry prostate volumes was investigated. Linear regression analysis was used to predict the TRAK (TRAKpred) using the prostate gland height, width and length (HWL) measurements. The TRAK was corrected using a conformal index (COIN) of the plan to improve the accuracy of the model. A further 54 patient plans were then used to evaluate the outcome of the linear regression lines.
A formula was proposed (HxWxLx蟺/5.19) to estimate the prostate volume, which showed a mean deviation from the planimetry volume of 0.0139 卤 3.339 cm3, where 69%was within 卤10%of planimetry, and 96%within 卤20%. Scatter graphs of TRAKTPS for 54 plans, showed a positive linear correlation, r = 0.98, using the planimetry volume, and r = 0.95 for the estimated volume. The further 54 treatment plans used for evaluation showed the TRAKpred was accurately predicted to 卤10%for 100%of plans using the planimetry volume and 81%using the prostate volume estimate.
The verification method proposed was found to be accurate and independent of TPS parameters, with suitable acceptance margins of 卤10%.