Minimizing static intensity modulation delivery time using an intensity solid paradigm

Abstract

Purpose: A leaf sequencing optimization algorithm that minimizes the delivery time for a static intensity modulated field is presented.

Methods and Materials: Sets of segments are created by intensity map operations subject to leaf collision constraints and tongue and groove effects. Each set’s delivery time is evaluated as a function of leaf travel, beam on time, and the verify and record (V&R) overhead. The configuration with the minimum delivery time is chosen. As a test, optimization was done on three clinical cases of varying complexity.

Results: Assuming 10 × 10-cm fields with an average of 17 intensity levels, the optimization technique reduced delivery times by 27% and 45%, when compared to rod pushing and power of two extraction, respectively. The treatment time for the optimal case with a V&R overhead of 4 s would be 11.5 min for 9 coplanar ports. Tongue-and-groove underdosages are removed, and the worst case leakage is 2% of the peak dose.

Conclusion: Compared to previously reported leaf sequencing methods, the new optimization algorithm described here reduces treatment times for complex static intensity modulated fields. Additionally, leakage is minimal and no tongue-and-groove underdosage occurs.

Introduction

Intensity modulation is a powerful tool for creating dose distributions that spare critical organs and that have high dose regions conforming to the tumor shape. This technique improves dose conformity by allowing the intensity of the beam to vary across the patient surface. This variation can be visualized as an intensity map (Fig.1a ), a 2-dimensional matrix whose cells have a one-to-one correspondence with the patient coordinates in the isocentric plane and whose entries are directly proportional to the amount of time that the corresponding cell of the matrix is exposed to radiation. The entries in the map are all integers, and a scaling factor multiplied to the map is used to determine the number of monitor units each cell receives.

Dose optimization routines 1, 2, 3 generate the intensity maps, and these maps could be delivered with techniques which involve compensators 2, 4, 5, dynamic multileaf collimation 6, 7, 8, 9, 10, 11, or static multileaf collimation 4, 10, 12, 13. For dynamic methods, many algorithms have been developed to generate the appropriate leaf sequences with the goal of minimizing the beam on time 6, 7, 8, 9, 10, 11. This goal is equivalent to minimizing the total treatment time, since the beam is on while the leaves are moving. The dynamic leaf sequencing concepts have been extended to static methods (10) by choosing appropriate intervals of the dynamic method and delivering the leaf settings in a stop-and-shoot mode. These methods also minimize the beam on time, but they do not minimize the total treatment time.

For static methods, every single segment can be verified and recorded, depending on the ability to prevent radiation from turning on, should there be any mismatch between the desired and the actual positions. However, there is an overhead associated with the confidence provided by the verification and record (V&R) cycle for every single leaf position of each segment. Since several segments are superimposed to generate the intensity map, this V&R overhead needs to be considered in the algorithms that generate the segments. Different sets of segments will also have different total beam on times and different amounts of leaf travel. These system related quantities must also be considered in the process of minimizing the total treatment time, and they must be treated in a way that accommodates changes in leaf speed, dose rate, and V&R overhead. For a system with a long V&R overhead, minimizing the number of segments may produce the minimum treatment time. If the machine has a low dose rate and a short V&R overhead, however, it is possible to find a set of segments which does not have the minimum number of segments, and yet has the minimum treatment time.

In this paper, a recently patented (14) optimization algorithm that minimizes the total delivery time by taking all the system parameters for a Siemens multi-leaf collimator (MLC) into account in a flexible way will be presented.