Z Jiang

An examination of the number of required apertures for step-and-shoot IMRT

We have examined the degree to which step-and-shoot IMRT treatment plans can be simplified (using a small number of apertures) without sacrificing the dosimetric quality of the plans. A key element of this study was the use of direct aperture optimization (DAO), an inverse planning technique where all of the multi-leaf collimator constraints are incorporated into the optimization. For seven cases (1 phantom, 1 prostate, 3 head-and-neck and 2 lung), DAO was used to perform a series of optimizations where the number of apertures per beam direction varied from 1 to 15. In this work, we attempt to provide general guidelines for how many apertures per beam direction are sufficient for various clinical cases using DAO. Analysis of the optimized treatment plans reveals that for most cases, only modest improvements in the objective function and the corresponding DVHs are seen beyond 5 apertures per beam direction. However, for more complex cases, some dosimetric gain can be achieved by increasing the number of apertures per beam direction beyond 5. Even in these cases, however, only modest improvements are observed beyond 9 apertures per beam direction. In our clinical experience, 38 out of the first 40 patients treated using IMRT plans produced using DAO were treated with 9 or fewer apertures per beam direction. The results indicate that many step-and-shoot IMRT treatment plans delivered today are more complex than necessary and can be simplified without sacrificing plan quality.

Effect of beamlet step-size on IMRT plan quality.

We have studied the degree to which beamlet step-size impacts the quality of intensity modulated radiation therapy (IMRT) treatment plans. Treatment planning for IMRT begins with the application of a grid that divides each beams-eye-view of the target into a number of smaller beamlets (pencil beams) of radiation. The total dose is computed as a weighted sum of the dose delivered by the individual beamlets. The width of each beamlet is set to match the width of the corresponding leaf of the multileaf collimator (MLC). The length of each beamlet (beamlet step-size) is parallel to the direction of leaf travel. The beamlet step-size represents the minimum stepping distance of the leaves of the MLC and is typically predetermined by the treatment planning system. This selection imposes an artificial constraint because the leaves of the MLC and the jaws can both move continuously. Removing the constraint can potentially improve the IMRT plan quality. In this study, the optimized results were achieved using an aperture-based inverse planning technique called direct aperture optimization (DAO). We have tested the relationship between pencil beam step-size and plan quality using the American College of Radiologys IMRT test case. For this case, a series of IMRT treatment plans were produced using beamlet step-sizes of 1, 2, 5, and 10 mm. Continuous improvements were seen with each reduction in beamlet step size. The maximum dose to the planning target volume (PTV) was reduced from 134.7% to 121.5% and the mean dose to the organ at risk (OAR) was reduced from 38.5% to 28.2% as the beamlet step-size was reduced from 10 to 1 mm. The smaller pencil beam sizes also led to steeper dose gradients at the junction between the target and the critical structure with gradients of 6.0, 7.6, 8.7, and 9.1 dose%/mm achieved for beamlet step sizes of 10, 5, 2, and 1 mm, respectively.

Direct aperture optimization of breast IMRT and the dosimetric impact of respiration motion

We have studied the application of direct aperture optimization (DAO) as an inverse planning tool for breast IMRT. Additionally, we have analysed the impact of respiratory motion on the quality of the delivered dose distribution. From this analysis, we have developed guidelines for balancing the desire for a high-quality optimized plan with the need to create a plan that will not degrade significantly in the presence of respiratory motion. For a DAO optimized breast IMRT plan, the tangential fields incorporate a flash field to cover the range of respiratory motion. The inverse planning algorithm then optimizes the shapes and weights of additional segments that are delivered in combination with the open fields. IMRT plans were generated using DAO with the relative weights of the open segments varied from 0% to 95%. To assess the impact of breathing motion, the dose distribution for the optimized IMRT plan was recalculated with the isocentre sampled from a predefined distribution in a Monte Carlo convolution/superposition dose engine with the breast simulated as a rigid object. The motion amplitudes applied in this study ranged from 0.5 to 2.0 cm. For a range of weighting levels assigned to the open field, comparisons were made between the static plans and the plans recalculated with motion. For the static plans, we found that uniform dose distributions could be generated with relative weights for the open segments equal to and below 80% and unacceptable levels of underdosage were observed with the weights larger than 80%. When simulated breathing motion was incorporated into the dose calculation, we observed a loss in dose uniformity as the weight of the open field was decreased to below 65%. More quantitatively, for each 1% decrease in the weight, the per cent volume of the target covered by at least 95% of the prescribed dose decreased by approximately 0.10% and 0.16% for motion amplitudes equal to 1.5 cm and 2.0 cm, respectively. When taking into account the motion effects, the most uniform and conformal dose distributions were achieved when the open segment weights were in the range of 65-80%. Within this range, high-quality IMRT plans were produced for each case. The study demonstrates that DAO with tangential fields provides a robust and efficient technique for breast IMRT planning and delivery when the open segment weight is selected between 65% and 80%.

Sampling issues for optimization in radiotherapy

A wide variety of optimization problems and techniques are used in radiation treatment planning. The problems typically involve large amounts of data, derived from simulations of patient anatomy and the properties of the delivery device. We investigate a three phase approach for their solution based on sampling of the underlying data that determines optimal beam angles, wedge orientations and delivery intensities in patient examples. Phase I uses multiple coarse samplings of the data and linear programming to adapt the sampling and determine a collection of promising angles to use. Phase II solves the adapted sample problems as mixed integer programs using only the promising angles. Phase III refines the sampling further, and fixes most of the discrete decision variables to reduce computation times. Particular emphasis will be given to general principles that are applicable to large classes of treatment planning problems. Specific examples show enormous increase in speed of planning, without detriment to the solution quality.

WE‐C‐M100F‐06: Dosimetric Comparison of High‐Z Inhomogeneity in IMRT: A Collaborative Study

With increased longevity, more prostate cancer patients with hip prosthesis are anticipated. Prosthetic devices have high atomic numbers (Z) and produce dose perturbation that is dependent on Z, beam energy, and depth. TG‐63 provided recommendations for high‐Z prosthetic devices in 3D conformal therapy. In IMRT, the dosimetric implication of the high‐Z prosthesis remains uncertain which is investigated in this collaborative study with 10 different treatment planning systems (TPS). Planning target volume (PTV), and the organs at risk (OAR) namely, the bladder, the rectum and a bilateral titanium hip‐prosthesis were contoured on a CT data set of a patient and sent to each collaborator with proper guidelines for beam arrangements, energy and dose volume constraints for planning. Due to significant streaking artifacts in the CT data, users were encouraged to use their own method to correct for redistribution of CT numbers, and assign the appropriate electron densities. Since dose perturbation is significant for low energy and less sensitive with multiple fields, equally distributed 7‐fields were planned. Beam energy was also studied for comparison. One common constraint, 95% PTV must receive at least 95% of dose was strictly followed by each planner. A variety of dose algorithms were used in different TPS, such as pencil beam, superposition, and convolution. Although the results of some planning systems are closer to each other, in general, there is a wide variation in dose distribution in PTV and the OARs, as well as the minimum, the maximum and the median doses which are commonly used in plan evaluation. The variation in MU and the number of segments also vary significantly. High energy beam provided slightly better but not significant dose distribution. Ranking of TPS cannot be established based on a single clinical case. A well‐controlled phantom study is planned to validate the merit of each TPS.

SU‐FF‐T‐101: Clinical Feasibility of “jaws‐Only” IMRT Using Direct Aperture Optimization

Purpose: To demonstrate the clinical feasibility of delivering IMRT treatment plans using only independent collimators.Method and Materials: The Direct Aperture Optimization (DAO) technique is used to optimize the jaw positions and the relative weights assigned to each aperture. Since all of the delivery constraints imposed by the jaws are incorporated into the optimization, the need for leaf sequencing is eliminated. This allows for “jaws‐only” IMRT plans with a significantly reduced number of segments as compared with “jaws‐only” plans produced with the traditional two‐step IMRT approach. We applied the DAO “jaws‐only” technique to three clinical cases: an abdomen, a prostate, and a head and neck. For each case, “jaws‐only” DAO (JODAO) plans were produced with 5, 10, 15, 20, and 25 apertures. For comparison, a DAO plan was created that utilized an MLC (MLCDAO). The resulting JODAO and MLCDAO plans were delivered to a phantom using an Elekta Precise linear accelerator.Results: The results demonstrate that between 15 and 25 “jaws‐only” apertures are required per beam direction to obtain conformal IMRT treatment plans that are comparable to the MLCDAO plans. The delivery times for the JODAO plans were between 15 and 20 minutes. This compares to the delivery times of 7 to 12 minutes for the MLCDAO plans. Conclusion: Using DAO, it is possible to create IMRT treatment plans that utilize only independent collimators. In addition, these “jaws‐only” plans can be delivered in a reasonable amount of time. This can make IMRT feasible in clinics which have linear accelerators not equipped with an MLC.

TH‐D‐AUD B‐06: Variability of Low‐Z Inhomogeneity Correction in IMRT/SBRT: A Multi‐Institutional Collaborative Study

Purpose: The dosimetry of IMRT beamlets with low‐Z inhomogeneities is a difficult problem and its accuracy is highly uncertain due to lateral disequilibrium. Various inhomogeneity correction algorithms: pencil beam (PB), collapsed cone convolution (CC), anisotropic analytical algorithm (AAA), Monte Carlo(MC), and combination of them are employed in different treatment planning systems (TPS) for dose calculations. This multi‐center collaborative study evaluates the accuracy and suitability of these algorithms for inhomogeneity correction in clinical trails. Method and Materials: A simple lung phantom was constructed with cork sheets (0.25 g/cm3) sandwiched between two 3 cm thick solid water slabs. The CT data of this phantom was sent to 8 institutions employing different TPS. Dose calculations were carried out at various depths over the field sizes: 0.5×0.5–10×10 cm2 for 6 and 15 MV beams with grid size (2×2 mm2). The calculated inhomogeneity correction factor (CF) was compared with measured data using a micro‐chamber. Results: The calculated CF with various algorithms showed marked variability and can be categorized in two classes; pencil beam (PB) and MC based collapsed cone (CC). The CF calculated with PB rises steadily in the lungtissue whereas CC exhibits the effect of electron transport. The differences between measured and calculated CF values varied from 70% to −10% for 6 MV from small to large fields. For 15 MV beam, the differences are even larger for small fields but reduce significantly for large field sizes (>5×5cm2). Conclusion: It is concluded that PB based algorithms should be avoided for dose calculation in small fields with low‐Z inhomogeneities. The MC derived kernel based algorithms such as CC and AAA produce similar results to each other within ±10% and should be preferred for patient treatment in IMRT/SBRT. This study raises serious concerns in dosimetric variability in lung cancers possibly impacting the clinical trials.

MO‐D‐M100J‐06: Reducing Intra‐Fraction Organ Motion Effects Using Segment Size Constraint in Direct Aperture Optimization

Purpose: In IMRT delivery, an important issue is intra‐fraction organ motion, which causes significant degradation of the delivered dose. Some simulation researches have showed that organ motion effects could be significantly reduced by increasing the segment size. In this study, a direct aperture optimization based commercial inverse planning system was modified to assess the clinical impact of creating optimized plans with segment size constraints (SSC). Our study seeks to answer what price in static plan quality one has to pay to avoid dose degradation in moving targets? Method and Materials:IMRT plans with and without SSC were optimized for two abdominal cases using static CTimages. The MLC travel direction is aligned with the projected motion direction and SSC penalizes all MLC leaf openings smaller than twice the projected motion amplitude. Static plans were recalculated with a sinusoidal target motion with 1cm amplitude and 4second period. Results: In both cases with and without SSC, PTV volume covered by 95% of the prescribed dose (V95) was less than 1%, indicating that SSC had little effect on static plan quality. After incorporating target motion, V95 was improved by applying SSC, with the degree of improvement depending on the particular case. For case 1, V95 improved from 84.5% to 93.3% by adding SSC. Further study shows that dose coverage of peripheral PTV regions improved more significantly with SSC than central regions. In all plans, differences of the doses to the critical structures were within a few percent. By adding SSC, the treatment plan was more tolerant to target motion. Conclusions: Our study showed that when segmental IMRT plans were delivered to a moving target, SSC improves delivered dose conformity (V95) as much as 9% without significantly sacrificing static plan quality. Peripheral PTV shows more improvement in dose coverage with SSC than central regions.

SU‐FF‐T‐91: An IMRT Planning Technique for Head‐And‐Neck Cancers That Utilizes Direct Aperture Optimization

Purpose:IMRT can play an important role in the irradiation of head and neck tumors traditionally treated by lateral fields matched with an anterior supraclivicular field. However, due to the complex PTV geometry, these IMRT plans result in large numbers of segments leading to inefficient deliveries. We have developed an alternative IMRT planning technique utilizing Direct Aperture Optimization (DAO) to streamline the planning process and provide significant efficiency gains. Method and Materials: The process begins with the placement of traditional 3D conformal fields (laterals and anerior superclavicular). Next, the dose is calculated with this beam arrangement. The 90% isodose line is converted into a PTV with surrounding critical structures (e.g. spinal cord, parotid glands, posterior medial neck region) excluded from the PTV definition. The resulting PTV serves as the target for IMRT planning. For our planning technique, we have used the DAO algorithm in the Prowess Panther planning system. DAO plans generally result in significantly fewer segments as compared with those generated by traditional IMRT planning techniques. This is of critical importance since traditional IMRT plans for these cases have excessively long treatment times. Using DAO allows practical treatment times without sacrificing plan quality. Results: Fifteen patients were planned and treated with this technique. Seven equispaced beams were used in each. The objectives were PTV dose conformity and low dose to any avoidance regions. The spinal cord limit was 40Gy. For the DAO plans, treatment times ranged from 9 to 17 minutes on an Elekta SL20 acclerator. For corresponding plans produced using Pinnacle3, treatment times ranged from 30 to 45 minutes. Conclusions: An IMRT treatment technique for head and neck cancer has been devised. This technique removes field matching and allows the initial 50Gy to be delivered with a single plan. Using DAO provides significant reductions in treatment times.

SU‐FF‐T‐86: An Automatic Field‐Matching Technique to Treat Multiple Targets with a Single IMRT Plan

Purpose: To treat multiple targets with a single IMRT plan with automatic field matching and different sets of angles for each target. In the treatment of head‐and‐neck (HN) malignancies with IMRT for example, the traditional approach is to deliver 7–9 IMRT fields matched with a static half‐beam blocked supraclavicular field. However, significant cold and hot spots are frequently observed near the field junction. We have developed a technique to generate a single IMRT plan that eliminates the need for beam matching and reduces excess irradiation of normal tissue. Method and Materials: Direct aperture optimization (DAO) [1, 2] is an inverse planning technique where the MLC delivery constraints are incorporated into the plan optimization. By defining the initial apertures prior to optimization, the IMRT fields are limited in the search space for the MLC leaves, which served as a seeding solution. The fields are restricted so as to prevent them from exceeding the beams eye view of their assigned targets. With this approach a single IMRT plan can be generated for multiple targets with different sets of gantry angles and automatic field matching. Results: Using DAO and defining the initial MLC aperture technique can produce a single IMRT plan for multiple targets without field matching. In the case of HN, 7 to 9 fields were assigned to the primary tumor, upper neck, and a portion of the lower neck nodes. An anterior and a posterior field were assigned to the mediastinum and a portion of the lower neck nodes. The resulting single isocenter IMRT plans were delivered without the need to junction fields. Conclusions: By using different beam arrangements, a single IMRT HN plan can be generated to treat multiple targets with needing to match fields.