Volume 34, Issue 6, June 2007
Index of content:
- Therapy Scientific Session: RoomM100J
- IMRT: Optimization
MO‐D‐M100J‐01: Dose Painting With Intensity Modulated Proton Therapy and Intensity Modulated X‐Ray Therapy: A Comparison34(2007); http://dx.doi.org/10.1118/1.2761242View Description Hide Description
Purpose: To compare intensity modulated proton therapy (IMPT) versus intensity modulated x‐ray therapy (IMXT) for the delivery of nonuniform dose prescriptions based on hypoxia‐imaging, so‐called dose painting. Materials and Methods: IMXT delivered with helical tomotherapy (HT) was compared to IMPT delivered with spot scanning (SS) and distal gradient tracking (DGT). The novel DGT method places beam spots where dose prescription gradients occur along the pencil beam axis. Fundamental dosimetric properties of each modality were assessed by creating optimized plans for 144 variations of a cylindrical phantom with six boost regions embedded inside a base tumor region. Clinical cases with biologically conformal dose prescriptions based on PET with the ‐ATSM hypoxia imagingradiopharmaceutical were planned. The effects on the nonuniform dose distribution of delivering IMPT on a 180° arc versus equi‐spaced beams spread over 360° were investigated. Results: Phantom studies showed that nonuniform dose plan quality for tomotherapy, SS, and DGT, was similar, but DGT plans were most sensitive to phantom size and boost region proximity. IMPT reduced normal tissue integral dose by a mean factor of around two relative to IMXT. Clinical dose deviations from the prescription were comparable for all modalities, but arc IMPT deliveries markedly reduced normal tissue dose and improved critical structure sparing without compromising the dose distribution in the tumor.Conclusions: In the target volume, IMXT and IMPT deliver comparable nonuniform dose distributions. IMPT offers improved integral normal tissue dose and sparing of critical structures over IMXT, as was the case for uniform dose deliveries. DGT reduces required beam spots by a factor of about three relative to SS. IMPT dose painting will require similar management of intrafractional patient motion as for IMXT, with the additional consideration of proton spot placement uncertainty.
TR Mackie has a conflict of interest due to financial interest in TomoTherapy, Inc.
34(2007); http://dx.doi.org/10.1118/1.2761243View Description Hide Description
Purpose: To develop a time efficient IMRTdelivery platform that simultaneously exploits all mechanical degrees of freedom of the linac.Method and Materials: Trajectory Based Radiation Therapy (TBRT) is a new technique for planning and delivering optimized dose distributions where the radiationsource moves along a continuous 3‐dimensional trajectory defined by gantry angle, couch angle, and couch position. The trajectory is constructed using a series of control points distributed along the trajectory. For planning, continuous source motion is modeled as a series of static beams with one beam defined at each control point. Highly restrictive constraints are placed on MLC and source motion to preserve a continuous, efficient and accurate delivery. Normally these restrictions would also severely limit the ability of the optimization algorithm to derive a high quality plan. This problem is solved using a novel technique for aperture based optimization where a coarse sampling of unrestricted control points is used in the initial stages of optimization. As the optimization progresses additional control points are added with increasing restrictions on MLC and source motion. This approach maintains time efficiency and delivery accuracy while allowing the optimization to derive a high quality plan. Results: Time studies have shown that TBRT delivery times are reduced to ∼ 1.5 to 3 minutes for a 200 cGy fraction. Thus far, results have shown that treatment plans generated with TBRT have dose distributions that are equivalent to or superior to static gantry IMRT.Conclusion: On‐line imaging techniques have provided clinicians with tools for verifying patient position and adapting treatment plans but at the expense of increased treatment time. TBRT is well suited for on‐line verification and adaptation with delivery times that are substantially shorter than static gantry IMRT, IMAT and Tomotherapy. Conflict of Interest: Supported in part by Varian Medical Systems.
MO‐D‐M100J‐03: Reducing the Sensitivity of IMPT Treatment Plans to Setup Errors and Range Variations34(2007); http://dx.doi.org/10.1118/1.2761244View Description Hide Description
Purpose: Intensity modulated proton therapy (IMPT) has the ability to deliver highly conformal dose distributions to tumors of complex shape. However, the accuracy of IMPT treatment plans is potentially compromised by various uncertainties, including setup errors and variations of the range of a proton beam in the patient. Method and Materials: We present two treatment planning concepts for IMPT which incorporate uncertainties into the optimization. The first approach minimizes the expectation value of the deviation of delivered dose and prescribed dose, assuming the delivered dose depends on several random variables that model the uncertainty. The second approach minimizes the maximal dose deviation that can occur. Results: It is shown that these methods can utilize the physical characteristics of the proton beam to make treatment plans relatively insensitive to a particular type of uncertainty.
A treatment plan optimized while accounting for range uncertainties avoids placing a bragg peak directly in front of the OAR. Instead, the lateral fall‐off of the pencil beam is utilized in order to avoid the risk of overdosing the OAR. In addition, the dose distributions delivered by individual beams are relatively homogeneous in beam direction. This ensures tumor coverage and homogeneity.
On the other hand, treatment plans optimized by accounting for setup errors (modeled as a rigid translation of the entire patient) utilize the distal fall‐off at a transition of OAR and tumor. In addition, the dose distributions delivered by individual beams avoid steep dose gradients perpendicular to the beam direction to ensure tumor homogeneity.
Robustness of a treatment plan with respect to both setup error and range variations is possible to a limited degree only, since both types or uncertainty favor conflictive treatment plans. Conclusion: The presented methods are valuable to make IMPT treatment plans insensitive to various types of uncertainty.
34(2007); http://dx.doi.org/10.1118/1.2761245View Description Hide Description
Purpose: To develop an IMRT optimization strategy in which CTV‐to‐PTV margins are iteratively reduced until the dosimetric margin distribution (DMD) meets a 90% population coverage criterion (PCC), with coverage being the percent of patients for which the CTV minimum dose is 100% or more of the planned PTV minimum dose. Method and Materials: The DMD is the 3D margin distribution between the CTV and the treated volume (TV). Consistent with ICRU guidelines, the TV is the volume enclosed by the PTV minimum dose isodose surface. Optimization utilizes an iterative approach which converges on a DMD that exactly meets a 90% PCC. In each iteration, the DMD is obtained by exporting the CTV and TV as meshes, and calculating the margins between them via a computational geometry package. Coverage with respect to the PCC is computed based upon the DMD. In the next iteration, the PTV is re‐created using a CTV‐to‐PTV margin determined via linear interpolation from prior coverage estimates to move the plan towards the desired PCC. Intensities are then reoptimized, and coverage is re‐evaluated. Iterations continue until the 90% PCC is met. Results: The proposed algorithm was run for a prostate IMRT plan. For 2mm simulated random and systematic setup errors, the algorithm yielded 3mm CTV‐to‐PTV margins, less than the 6.4mm margin prescribed by the van Herk margin formula (VHMF). While the DMD maintained the desired 90% coverage, the TV was reduced by 63% relative to a VHMF CTV‐to‐PTV margin plan. Conclusion: To accommodate patient setup errors in a self‐consistent manner, it is feasible and desirable to generate IMRT plans in which the DMD meets a population coverage criterion. DMDs assure the desired PCC is achieved and can result in lower treated volumes compared with traditional margin approaches. (Work supported by NIH R01CA98524).
34(2007); http://dx.doi.org/10.1118/1.2761247View Description Hide Description
Uncertainties in the precise locations of tumor voxels are inherent for a variety of reasons. It is therefore desirable to take those uncertainties into account, and further, to derive treatment planning methodologies that directly utilize information (e.g. probability distributions or actual trajectories) about the location of the target, including spatial uncertainties, so that resulting treatment plans exhibit normal tissue dose reduction, and potentially higher tumor doses.
Two novel robust approaches are investigated. 1)Under‐dose probability(UDP) estimates the probability of the tumor receiving a dose less than a user‐defined critical dose. It can be modeled using linear mixed‐integer programming techniques. 2)Probability dose generation(PDG) designs treatment plans in which the probability of a voxel located in a target is factored during the dose generation process. Utilizing 4D‐CT scans of lung/liver cancer patients during different breathing phases(phases 0–9, 0:full‐inhale, 5:full‐exhale), four treatment planning strategies are compared. 1)Standard planning with a static PTV based on a single selected phase (control). 2)The Internal Target Volume (ITV) approach, where ITV is defined as the union of CTVs in all breathing phases. 3)Plans obtained with UDP. 4)Plans obtained via PDG. Sophisticated computational optimization techniques are used to solve each of these models.
Compared to Standard and ITV plans, UDP and PDG plans offer good coverage, comparable min‐PTV‐dose, improved PTV‐conformity, and higher PTV‐dose. In addition, in lung, PDG reduces normal‐lung‐mean‐dose by 13% and heart‐mean‐dose by 20%; and UDP reduces esophagus‐mean‐dose by 70% and heart‐mean‐dose by 49%. In liver, PDG reduces normal‐liver‐ mean‐dose by 8% and other‐normal‐tissue‐mean‐dose by 26%, whereas UDP reduces normal‐liver‐mean‐dose by 10%, both with improved PTV‐conformity of 7%.
Thus, the UDP and PDG approaches can result in treatment plans of increased robustness, they allow a reduction in mean dose to organs‐at‐risk and normal tissue, and in some cases, deliver higher dose to the tumor volume.
MO‐D‐M100J‐06: Reducing Intra‐Fraction Organ Motion Effects Using Segment Size Constraint in Direct Aperture Optimization34(2007); http://dx.doi.org/10.1118/1.2761248View Description Hide Description
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.
34(2007); http://dx.doi.org/10.1118/1.2761249View Description Hide Description
Purpose: The conventional treatment planning paradigm involves a plan‐and‐evaluate followed by a modify‐if‐necessary approach. We describe an IMRTtreatment planning framework in which multiple plans that differ in the input constraints are generated per case prior to evaluation. We also describe a decision‐support‐system (DSS) for evaluating and ranking multiple plans and hypothesize that the planning surface can be modeled using quadratic modeling. Methods and Materials: One hundred twenty‐five plans were generated sequentially for a head‐and‐neck case and a pelvic case by varying the dose‐volume constraints on each of the OARs. A DSS was used to rank plans according to DVH and equivalent uniform dose (EUD) values using composite criteria and pre‐emptive selection. Two methods for ranking treatment plans were evaluated: composite criteria and pre‐emptive selection. The planning surfaces corresponding to the 125‐plan sets were modeled using quadratic functions by formulating the problem as a linear program. Results: The DSS provides an interface for the comparison of multiple plan features. Plan ranking using both composite and pre‐emptive criteria resulted in the reduction of plan space to 1–3 “optimal” plans. The planning surface models had good predictive capability with respect to both DVH and EUD values with fit errors of < 6%. Models generated by minimizing the maximum relative error had significantly lower relative errors than models obtained by minimizing the sum of squared errors. The inter‐dependence of OAR‐OAR properties could be successfully inferred for both clinical cases through the use of contour plots, which represent projections through the multi‐dimensional plan property space. Conclusion: The DSS can be used to aid the planner in the selection of the most desirable plan. The collection of quadratic models constructed from the plan data in order to predict DVH and EUD values generally showed excellent agreement with the actual plan values.
34(2007); http://dx.doi.org/10.1118/1.2761250View Description Hide Description
Purpose: The feasibility of delivering deliberately heterogeneous dose distributions based on biological image data is related to transformations that map the image data to dose prescription values, which is currently unknown. We created optimized dose distributions that were based on four possible image‐to‐dose‐prescription transformations to determine their delivery feasibility. Materials and Methods: Tomotherapy photon dose calculations were performed for a canine with a nasopharyngeal tumor using the Convolution/Superposition algorithm. Linear, square root, quadratic, and Gompertz transformations between the standard uptake values (SUV) from ‐ATSM PETimages and the prescriptions were implemented to yield weighted distributions of prescribed dose boosts in regions of expected radio‐resistant hypoxic cells. Maximum dose boosts were constrained to reflect clinically realistic whole tumor doses and constant normal tissue doses. Optimized heterogeneous dose distributions were found by minimizing a voxel‐by‐voxel quadratic objective function in which all tumor voxels were given equal importance weightings. Results: The planned non‐uniform dose distributions founded on hypoxia maps display variability between preservation and resolution of biological heterogeneities. Whereas the linear transform preserved the SUV distribution with similar heterogeneities in the prescription, the square root and quadratic transforms softened or sharpened heterogeneities in respective prescriptions. The square root dose plan matched its prescription well due to shallow dose gradients but failed to resemble the hypoxia map, while the quadratic plan poorly matched its prescription due to steep gradients but resembled the PET scan. The linear planned dose distribution matched both its prescription and the biological image fairly well. Conclusions: Simple transformations investigate the importance of knowing given prescriptions exactly to deliver non‐uniform dose distributions from biological images. Transformations that blur heterogeneities lead to dose distributions that match their prescriptions coupled with loss of conformality to the PETimages. Those preserving heterogeneities lead to less well‐matched distributions but do conform to the scans.