The problem of template-based inverse planning in brachytherapy can be formulated as follows:

Given a template which predefines all possible catheter positions and directions, find the catheters and source dwell positions such that the resulting dose distribution is optimal.

Ideally, the optimal solution (i.e. catheter and source dwell configuration) should be found by starting from the desired dose distribution and inverse-engineering the configuration that results to that distribution. However, this approach has been proven unsatisfactory [1].

A practical solution is Multiobjective Optimization, which is succesfully used in conventional dose optimization for years. The problem of dose optimization involves objective functions such as the maximization of the PTV (Planning Target Volume) coverage and the minimization of dose values in surrounding normal tissue. The task to satisfy both kinds of objectives simultaneously is not trivial, and multiobjective optimization has proven to be successful.

The basic difference of Inverse Planning is that the location of each catheter is not known ad hoc, and another class of objectives has to be introduced: the position and the number of catheters. Typically, the number of catheters must be kept as low as possible, and they should avoid inflicting damage to nearby organs. This additional objective causes a massive expansion of the search space, since the possible source positions, which define the dose distribution, depend on the positions of the catheters. Other necessary tasks, such as the calculation of the possible source dwell positions and automatic source activation [3] can be performed efficiently in a preprocessing stage.

Two different approaches to the problem have been investigated, one which uses gradient-based deterministic optimization algorithms where the objectives are represented as a single weighted sum, and one true multiobjective approach which uses stochastic search algorithms [2]. However, in order to be used efficiently by a treatment planner, a moderate level of expertise in the field of multiobjective optimization is required.

### Hybrid Inverse Planning and Optimization (HIPO)

HIPO is an inverse planning tool, designed and developed by Pi-Medical [4]. This tool utilizes topology related techniques and a stochastic algorithm for adjusting and adapting the catheter/needle configuration, where the adjustment of dwell times of the source dwell positions within the catheters is realized using the Dose Volume Histogram Optimization (DVHO) algorithm.

Firstly HIPO finds out all catheters, the so called feasible catheters, and corresponding template holes, that can be inserted based on the Placement Settings defined by the user. Thereafter HIPO begins to search for the best placement of the user-defined number of catheters, based on the total number of feasible catheters and the objectives and penalties defined by the user.

Attachment | Size |
---|---|

HIPO: A hybrid inverse treatment planning optimization algorithm in HDR Brachytherapy | 1.24 MB |