Abstract:To quantitatively compare the aperture-based and beamlet-based inverse planning techniques and draw conclusion with regard to the dependence of plan quality on the number of segments of the incident fields using three clinical cases of different disease sites. We developed a combined aperture based inverse planning algorithm. After the number of segments was specified, the beam eyes views (BEVs) are applied to delineate the maximum MLC field boundary for each gantry angle. The BEV apertures were chosen to be the initial settings of aperture shapes and served as physical constraints of aperture settings. In the proposed algorithm, we utilized the fast adaptive simulated annealing (ASA) technique to select the field shapes defined by the MLC leaf coordinates. A descent gradient-based optimization technique was employed to search for the optimal aperture weights under the guidance of a conventional quadratic objective function. By combing the fast ASA with the descent gradient search algorithm, the computational efficiency was consistently improved. The beamlet-based inverse planning used the same objective function and the descent gradient-based optimization. Beamlet-based optimization with 1cm × 1cm size was used to obtain the optimized fluence maps (10 intensity levels). Three clinical cases (prostate, thoracic, and head and neck) were selected for the comparative study. For each case, the gantry angles are chosen manually and beamlet-based inverse planning was performed. Aperture based optimization was done with the same beam configuration and beam modality. A range of number of segments (5 to 25) was investigated. Plans with different number of segments were evaluated based on the DVH and dose distribution improvements and the treatment times involved. The combined aperture-based inverse planning algorithm was implemented and tested by comparing with conventional simulated annealing algorithm for aperture-based IMRT. The same quality plans were achieved but the computation time was reduced even by an order of magnitude. The improvement in performance directly was directly resulted from the fact that the objective function is well behaved analytical function with respect to beam weights. The key results found from the comparative study are that the dose distribution and DVHs were improved with the increase of the number of segments. When the number is increased up to ∼14, the dose distribution obtained from aperture based optimization is quite similar to that of the beamlet-based optimization provided the leaves can be adjusted in 0.2cm increments in the aperture-based inverse planning. Further increase of the segment number only lead to minor improvement in dose distributions with much increased computing time. Considering the tradeoff between the plan quality and planning time, it seems that the optimal number of segment is between 10 ∼ 20. Aperture-based inverse planning seems to be a viable option for IMRT and eliminating the step of leaf sequencing. For a few different disease sites, our study indicate that the technique can produce IMRT plans comparable with that of beamlet-based inverse planning when the number of segments per fields is above ∼10. The improvement saturates gradually as the number of segments increased to above ∼20.

Prior Knowledge Helps Improve Beam Angle Optimization Efficiency in Radiotherapy Planning

Quantitative Comparison of Aperture-Based and Beamlet-Based Inverse Planning Techniques

A Surrogate-Based Metaheuristic Global Search Method for Beam Angle Selection in Radiation Treatment Planning

Optimization Strategies for Beam Direction and Dose Distribution Selection in Radiotherapy Planning

A genetic algorithm with neural network fitness function evaluation for IMRT beam angle optimization

Deep Reinforcement Learning for Beam Angle Optimization of Intensity-Modulated Radiation Therapy

Towards the optimization of ballistics in proton therapy using genetic algorithms: implementation issues

MO-D-351-03: A Unified Approach to Beam Angle Selection and Dose Optimization with High-Throughput Computing for IMRT

A reinforcement learning application of guided Monte Carlo Tree Search algorithm for beam orientation selection in radiation therapy

Multi-objective optimization in radiotherapy: applications to stereotactic radiosurgery and prostate brachytherapy

On polling directions for randomized direct-search approaches: application to beam angle optimization in intensity-modulated proton therapy

On-line Dose Calculation Using Deep Learning for Beams Selection in Non-Coplanar Radiotherapy

Two Effective Heuristics for Beam Angle Optimization in Radiation Therapy

Beam Orientation Optimization for Intensity-Modulated Radiation Therapy Using Mixed Integer Programming

Lung IMRT planning with automatic determination of beam angle configurations.

Metaheuristic Algorithm for Constrained Optimization in Radiation Therapy Treatment Planning: Design and Performance Comparison

Efficient optimization of an accelerator neutron source for neutron capture therapy using genetic algorithms

Variable length genetic algorithm with continuous parameters optimization of beam layout in proton therapy

Evaluating machine learning enhanced intelligent‐optimization‐engine (IOE) performance for ethos head‐and‐neck (HN) plan generation

Applying mixed-integer linear programming to the non-coplanar beam angle optimization of intensity-modulated radiotherapy for liver cancer

Knowledge Models as Teaching Aid for Training Intensity Modulated Radiation Therapy Planning: A Lung Cancer Case Study