Intraoperative optimization of needle placement and dwell times for conformal prostate brachytherapy.

PURPOSE: Traditionally, transperineal prostate brachytherapy has been heavily operator dependent. To overcome this limitation, a treatment planning method was developed for intraoperative planning, guidance, and evaluation. In this setting, reliability, speed, and ease of understanding are primary considerations. This planning method has been implemented for ultrasound guided implants of the prostate, but can be extended for use in other body sites. METHODS AND MATERIALS: The length and cross-section of the target (prostate) and location of urethra and rectum are determined intraoperatively from live ultrasound imaging. The planning program then automatically generates a "reference plan" containing needle locations, dwell times, and the resulting isodose distribution. As needles are placed, this information is corrected to account for any deviation of needle placement or movement of the prostate. Once all needles are in place, the normalization is adjusted to reconcile remaining hot-spots with coverage of the target volume. Optimization is performed in three separate stages. Each stage works to enhance only a subset of the implant parameters. (a) Pattern Optimization attempts to find the most appropriate placement for the needles or catheters. It is based on the transverse contour of the target volume. Needles are placed uniformly around the perimeter, and interior needle positions are determined from the cross-sectional area and shape. Critical structures such as the urethra are explicitly avoided. This step provides the overall framework for the implant, and is not generally repeated. (b) Relative Dwell Time Optimization selects relative dwell times that will give the best uniformity of dose. It works by setting the relative dwell time in each source position inversely proportional to the dose delivered to that point by the other source positions. It is used in the reference plan, and is repeated as each needle is inserted. This provides dosimetric feedback to the physician, who can judge the effect of deviations from the reference plan. (c) Relative Volume Optimization is an interactive method for fine tuning the normalization based upon volume analysis. The volume analysis is presented in tabular and graphical form, both being updated rapidly as the normalization is adjusted. The information is formatted to help the operator judge coverage and uniformity. Special functions are provided that allow the operator to "jump" to special normalization values based on several indices of uniformity or uniformity/coverage.
RESULTS: This system overcomes some conventional brachytherapy limitations. Rather than depend on the operator's intuitive judgement of where the needles should be placed, a global plan is generated and validated with full dose calculations. Immediate feedback is provided concerning the adequacy of placement and avoidance of critical structures. This information is provided in terms of actual tissue doses to the target volume and critical structures using point doses, isodose distributions, and volume analysis. Since the new method was introduced in January 1994, 33 implants have been performed. The needle placement method has been reliable in the clinic, with different doctors producing similar results on subsequent fractions for the same patient.
CONCLUSION: The method of decomposing the optimization problem into several simple steps is capable of rapidly, consistently, and reliably designing conformal treatment plans of high uniformity. Operator dependence has been significantly reduced. We are adapting the method for other anatomic sites.