- Email: [email protected]
Unbiased Metrics of Dosimetric Variation in IMRT Planning
S758
International Journal of Radiation Oncology Biology Physics
Conclusion: The parameters derived in this NTCP modeling study suggest that the TD50 for cardiac toxicity for female patients could be as much as 19Gy lower than for male patients during radiation therapy for esophageal cancer. Needing validation in an independent data set, this is the first study of its kind to suggest different dose volume constraints based upon gender. Author Disclosure: M. Snyder: None. J. Burmeister: None. M. Joiner: None. J. Meyer: None. L. Tait: None. S. Cohen: None. E. McSpadden: None. A. Konski: None.
a multi-objective evolutionary algorithm (MOEA) to generate spatially optimized dose distributions is proposed. An iterative dialog between the MOEA and a mathematical model that is used to predict tumor cell proliferation, diffusion, and response to radiation therapy generates weekly dose distributions that attempt to maximize tumor control and minimize normal tissue damage. Materials/Methods: An MOEA was used to optimize IMRT dose distributions for two patients throughout simulated treatment using a published mathematical model of glioblastoma proliferation and invasion to calculate tumor cell distributions. Optimal dose distributions were calculated by the MOEA using predictions of cellular distribution from the mathematical model for each week of treatment. The MOEA is a general optimization approach and can use any type of decision criteria to judge plan quality. Plans will approach the Pareto front that represents optimal trade offs amongst the clinical goals as defined by the decision criteria. The effectiveness of three tumor control decision criteria were compared: maximum tumor cell kill, minimum tumor cell survival after 1 week of treatment, and minimum tumor cell survival 12 weeks post treatment. Dose distributions were scaled so that the EUD to the normal tissue per fraction was less than the EUD delivered using a standard 1.8 Gy to the tumor with a 2.5 cm margin. Normal tissue EUDs and days gained, the time required to reach pretreatment tumor size as predicted by the mathematic model, were compared for optimized plans and standard IMRT plans. Results: Optimized dose distributions were characterized by higher dose to smaller volumes of tissue. When the maximum dose was restricted to 3 Gy, tumor control was similar to the standard plan, but normal tissue EUD was reduced from w1 Gy/fraction to w0.25 Gy/fraction for both patients. For one patient, all other MOEA IMRT plans resulted in w100 days gained versus only w20 days gained for the standard plan. For the second patient, the clinical plan resulted in w60 days gained. Optimized plans that restrict the maximum dose to 8 Gy and either maximized tumor cell kill or minimized tumor cell survival after 1 week of treatment resulted in w110 days gained (normal tissue EUD Z 0.55 Gy/fraction). Using an 8 Gy maximum dose and minimizing modeled tumor cell growth 12 weeks post treatment gave the best result of 150 days gained (normal tissue EUD Z 0.87 Gy/fraction). Conclusions: A patient-specific mathematical model showed that an adaptive IMRT treatment scheme using complex dose distributions could reduce normal tissue damage and increase tumor control. The MOEA is flexible and can be integrated with mathematical models of any type of cancer. Author Disclosure: C. Holdsworth: None. D. Corwin: None. R. Stewart: None. R. Rockne: None. A.D. Trister: None. K. Swanson: None. M. Phillips: None.
3435 Unbiased Metrics of Dosimetric Variation in IMRT Planning O. Nohadani,1 S. Srivastava,1 C. Medawar,1 and I.J. Das2; 1Purdue University, West Lafayette, IN, 2Indiana University Medical Center, Indianapolis, IN Purpose/Objective(s): The process of IMRT planning is an iterative inverse process, where a planner seeks to attain a desired dose distribution, specified for tumors and OARs, hence creating a plethora of competing constraints. However, the final product is rather the unpredictable outcome of a series of trial-and-error attempts at meeting these competing objectives. We seek to provide a set of metrics for unbiased DVH comparison for inter-planner variations in IMRT planning with identical dose prescription and optimization for estimation of quality of treatment plans. Materials/Methods: For one clinical prostate case, the treatment was planned by seven planners, with identical data set including beam parameters and target volumes while imposing the same clinical objectives for PTV and OARs, using a treatment planning system and accelerator. The resulting seven DVHs were compared based on their deviation of from the ideal coverage, namely 100% of PTV receiving 100% of the dose. Our deviation function L measures the area between the ideal and realized DVH. A weight function W is used for dose dependency. We employ a set of Ws: constant, piece-wise linear, quadratic, normal and beta-distributional weight functions to fulfill the goal. Results: There were large variations in the IMRT plans reflected by the DVH among planners even with identical dose-volume constraints and with same planning system. In fact, for a constant W, the inter-planner variations in PTV were as large as 20%. We show that the deviations of some planners are consistently higher than others for all eight Ws. Similarly, we observe that one of the planners consistently exhibits the lowest deviation, while another one is low for linear and quadratic Ws and worsen for the distributional functions independent of their slope and range. Further, the normal and beta-density function weights, as they primarily penalize the range of 95% < Dose < 105%, hence discriminating only deviations in the respective region. Conclusion: The outcome of treatment planning is strongly dependent on the planner’s skill and choices, hence variable in outcome. The proposed set of DVH metrics allow for unbiased comparison, beyond the visual inspections. An optimized superposition of these metrics can yield a practical tool for daily treatment planning. Furthermore, these results exhibit the need for methods whose outcomes are independent of the planning personal. This approach may also serve as an unbiased standard for future training and protocol design for medical physicists and treatment planners, as well as an additional quantitative measure for the daily clinical decision making. Author Disclosure: O. Nohadani: E. Research Grant; Indiana University / Purdue University joint grant. S. Srivastava: E. Research Grant; Indiana University / Purdue University joint grant. C. Medawar: E. Research Grant; Indiana University / Purdue University joint grant. I.J. Das: E. Research Grant; Indiana University / Purdue University joint grant.
3436 Adaptive IMRT Using a Multiobjective Evolutionary Algorithm Integrated With a Diffusion-Invasion Model for Glioblastoma C. Holdsworth, D. Corwin, R. Stewart, R. Rockne, A.D. Trister, K. Swanson, and M. Phillips; University of Washington Medical Center, Seattle, WA Purpose/Objective(s): A patient-specific method of adaptive intensity modulated radiation therapy (IMRT) treatment of glioblastoma using
3437 Observation of Radiation-induced Tissue Signal Intensity Changes With the First Commercial MRI-guided IMRT System O.L. Green, Y. Hu, C. Noel, J.R. Olsen, and S. Mutic; Washington University School of Medicine, St Louis, MO Purpose/Objective(s): To describe radiation-induced tissue signal intensity changes observable with a commercial MRI-Guided IMRT system and the potential utility of these images. The MRI-Guided IMRT system was recently installed at our institution. The stand-alone system is comprised of a 0.35-T double-doughnut MRI and a gantry housing three Co-60 sources that provide IMRT delivery with 600 cGy maximum dose rate at isocenter. Prior to installation of the sources, a patient imaging study was conducted to evaluate the imaging quality of this hybrid system. During the course of this imaging study, an observation was made of radiation-induced signal intensity change in the liver of a patient who had recently completed radiation therapy for pancreatic cancer. Previously published reports noted tissue changes visible on CT and 1.5-T diagnostic MRI systems; this is the first observation of radiation-induced tissue change with a low-strength MRI specifically designed for integration with a radiation delivery system. Materials/Methods: The patient’s radiation prescription was 5500 cGy in 25 fractions delivered via 6-MV IMRT; the scan was performed ten days
International Journal of Radiation Oncology Biology Physics
Conclusion: The parameters derived in this NTCP modeling study suggest that the TD50 for cardiac toxicity for female patients could be as much as 19Gy lower than for male patients during radiation therapy for esophageal cancer. Needing validation in an independent data set, this is the first study of its kind to suggest different dose volume constraints based upon gender. Author Disclosure: M. Snyder: None. J. Burmeister: None. M. Joiner: None. J. Meyer: None. L. Tait: None. S. Cohen: None. E. McSpadden: None. A. Konski: None.
a multi-objective evolutionary algorithm (MOEA) to generate spatially optimized dose distributions is proposed. An iterative dialog between the MOEA and a mathematical model that is used to predict tumor cell proliferation, diffusion, and response to radiation therapy generates weekly dose distributions that attempt to maximize tumor control and minimize normal tissue damage. Materials/Methods: An MOEA was used to optimize IMRT dose distributions for two patients throughout simulated treatment using a published mathematical model of glioblastoma proliferation and invasion to calculate tumor cell distributions. Optimal dose distributions were calculated by the MOEA using predictions of cellular distribution from the mathematical model for each week of treatment. The MOEA is a general optimization approach and can use any type of decision criteria to judge plan quality. Plans will approach the Pareto front that represents optimal trade offs amongst the clinical goals as defined by the decision criteria. The effectiveness of three tumor control decision criteria were compared: maximum tumor cell kill, minimum tumor cell survival after 1 week of treatment, and minimum tumor cell survival 12 weeks post treatment. Dose distributions were scaled so that the EUD to the normal tissue per fraction was less than the EUD delivered using a standard 1.8 Gy to the tumor with a 2.5 cm margin. Normal tissue EUDs and days gained, the time required to reach pretreatment tumor size as predicted by the mathematic model, were compared for optimized plans and standard IMRT plans. Results: Optimized dose distributions were characterized by higher dose to smaller volumes of tissue. When the maximum dose was restricted to 3 Gy, tumor control was similar to the standard plan, but normal tissue EUD was reduced from w1 Gy/fraction to w0.25 Gy/fraction for both patients. For one patient, all other MOEA IMRT plans resulted in w100 days gained versus only w20 days gained for the standard plan. For the second patient, the clinical plan resulted in w60 days gained. Optimized plans that restrict the maximum dose to 8 Gy and either maximized tumor cell kill or minimized tumor cell survival after 1 week of treatment resulted in w110 days gained (normal tissue EUD Z 0.55 Gy/fraction). Using an 8 Gy maximum dose and minimizing modeled tumor cell growth 12 weeks post treatment gave the best result of 150 days gained (normal tissue EUD Z 0.87 Gy/fraction). Conclusions: A patient-specific mathematical model showed that an adaptive IMRT treatment scheme using complex dose distributions could reduce normal tissue damage and increase tumor control. The MOEA is flexible and can be integrated with mathematical models of any type of cancer. Author Disclosure: C. Holdsworth: None. D. Corwin: None. R. Stewart: None. R. Rockne: None. A.D. Trister: None. K. Swanson: None. M. Phillips: None.
3435 Unbiased Metrics of Dosimetric Variation in IMRT Planning O. Nohadani,1 S. Srivastava,1 C. Medawar,1 and I.J. Das2; 1Purdue University, West Lafayette, IN, 2Indiana University Medical Center, Indianapolis, IN Purpose/Objective(s): The process of IMRT planning is an iterative inverse process, where a planner seeks to attain a desired dose distribution, specified for tumors and OARs, hence creating a plethora of competing constraints. However, the final product is rather the unpredictable outcome of a series of trial-and-error attempts at meeting these competing objectives. We seek to provide a set of metrics for unbiased DVH comparison for inter-planner variations in IMRT planning with identical dose prescription and optimization for estimation of quality of treatment plans. Materials/Methods: For one clinical prostate case, the treatment was planned by seven planners, with identical data set including beam parameters and target volumes while imposing the same clinical objectives for PTV and OARs, using a treatment planning system and accelerator. The resulting seven DVHs were compared based on their deviation of from the ideal coverage, namely 100% of PTV receiving 100% of the dose. Our deviation function L measures the area between the ideal and realized DVH. A weight function W is used for dose dependency. We employ a set of Ws: constant, piece-wise linear, quadratic, normal and beta-distributional weight functions to fulfill the goal. Results: There were large variations in the IMRT plans reflected by the DVH among planners even with identical dose-volume constraints and with same planning system. In fact, for a constant W, the inter-planner variations in PTV were as large as 20%. We show that the deviations of some planners are consistently higher than others for all eight Ws. Similarly, we observe that one of the planners consistently exhibits the lowest deviation, while another one is low for linear and quadratic Ws and worsen for the distributional functions independent of their slope and range. Further, the normal and beta-density function weights, as they primarily penalize the range of 95% < Dose < 105%, hence discriminating only deviations in the respective region. Conclusion: The outcome of treatment planning is strongly dependent on the planner’s skill and choices, hence variable in outcome. The proposed set of DVH metrics allow for unbiased comparison, beyond the visual inspections. An optimized superposition of these metrics can yield a practical tool for daily treatment planning. Furthermore, these results exhibit the need for methods whose outcomes are independent of the planning personal. This approach may also serve as an unbiased standard for future training and protocol design for medical physicists and treatment planners, as well as an additional quantitative measure for the daily clinical decision making. Author Disclosure: O. Nohadani: E. Research Grant; Indiana University / Purdue University joint grant. S. Srivastava: E. Research Grant; Indiana University / Purdue University joint grant. C. Medawar: E. Research Grant; Indiana University / Purdue University joint grant. I.J. Das: E. Research Grant; Indiana University / Purdue University joint grant.
3436 Adaptive IMRT Using a Multiobjective Evolutionary Algorithm Integrated With a Diffusion-Invasion Model for Glioblastoma C. Holdsworth, D. Corwin, R. Stewart, R. Rockne, A.D. Trister, K. Swanson, and M. Phillips; University of Washington Medical Center, Seattle, WA Purpose/Objective(s): A patient-specific method of adaptive intensity modulated radiation therapy (IMRT) treatment of glioblastoma using
3437 Observation of Radiation-induced Tissue Signal Intensity Changes With the First Commercial MRI-guided IMRT System O.L. Green, Y. Hu, C. Noel, J.R. Olsen, and S. Mutic; Washington University School of Medicine, St Louis, MO Purpose/Objective(s): To describe radiation-induced tissue signal intensity changes observable with a commercial MRI-Guided IMRT system and the potential utility of these images. The MRI-Guided IMRT system was recently installed at our institution. The stand-alone system is comprised of a 0.35-T double-doughnut MRI and a gantry housing three Co-60 sources that provide IMRT delivery with 600 cGy maximum dose rate at isocenter. Prior to installation of the sources, a patient imaging study was conducted to evaluate the imaging quality of this hybrid system. During the course of this imaging study, an observation was made of radiation-induced signal intensity change in the liver of a patient who had recently completed radiation therapy for pancreatic cancer. Previously published reports noted tissue changes visible on CT and 1.5-T diagnostic MRI systems; this is the first observation of radiation-induced tissue change with a low-strength MRI specifically designed for integration with a radiation delivery system. Materials/Methods: The patient’s radiation prescription was 5500 cGy in 25 fractions delivered via 6-MV IMRT; the scan was performed ten days