2020
Mao, Ximeng; Pineau, Joelle; Keyes, Roy; Enger, Shirin A.
RapidBrachyDL: Rapid Radiation Dose Calculations in Brachytherapy Via Deep Learning Journal Article
In: International Journal of Radiation Oncology, Biology, Physics, vol. 108, no. 3, pp. 802–812, 2020, ISSN: 1879-355X.
Abstract | Links | BibTeX | Tags: Brachytherapy, Colon, Computer, Computer-Assisted, Deep Learning, Female, Humans, Iridium Radioisotopes, Male, Monte Carlo Method, Neural Networks, Organs at Risk, Prostate, Prostatic Neoplasms, Radiotherapy Dosage, Radiotherapy Planning, Rectum, Retrospective Studies, Sigmoid, Urinary Bladder, Uterine Cervical Neoplasms
@article{mao_rapidbrachydl_2020,
title = {RapidBrachyDL: Rapid Radiation Dose Calculations in Brachytherapy Via Deep Learning},
author = {Ximeng Mao and Joelle Pineau and Roy Keyes and Shirin A. Enger},
doi = {10.1016/j.ijrobp.2020.04.045},
issn = {1879-355X},
year = {2020},
date = {2020-11-01},
journal = {International Journal of Radiation Oncology, Biology, Physics},
volume = {108},
number = {3},
pages = {802--812},
abstract = {PURPOSE: Detailed and accurate absorbed dose calculations from radiation interactions with the human body can be obtained with the Monte Carlo (MC) method. However, the MC method can be slow for use in the time-sensitive clinical workflow. The aim of this study was to provide a solution to the accuracy-time trade-off for 192Ir-based high-dose-rate brachytherapy by using deep learning.
METHODS AND MATERIALS: RapidBrachyDL, a 3-dimensional deep convolutional neural network (CNN) model, is proposed to predict dose distributions calculated with the MC method given a patient's computed tomography images, contours of clinical target volume (CTV) and organs at risk, and treatment plan. Sixty-one patients with prostate cancer and 10 patients with cervical cancer were included in this study, with data from 47 patients with prostate cancer being used to train the model.
RESULTS: Compared with ground truth MC simulations, the predicted dose distributions by RapidBrachyDL showed a consistent shape in the dose-volume histograms (DVHs); comparable DVH dosimetric indices including 0.73% difference for prostate CTV D90, 1.1% for rectum D2cc, 1.45% for urethra D0.1cc, and 1.05% for bladder D2cc; and substantially smaller prediction time, acceleration by a factor of 300. RapidBrachyDL also demonstrated good generalization to cervical data with 1.73%, 2.46%, 1.68%, and 1.74% difference for CTV D90, rectum D2cc, sigmoid D2cc, and bladder D2cc, respectively, which was unseen during the training.
CONCLUSION: Deep CNN-based dose estimation is a promising method for patient-specific brachytherapy dosimetry. Desired radiation quantities can be obtained with accuracies arbitrarily close to those of the source MC algorithm, but with much faster computation times. The idea behind deep CNN-based dose estimation can be safely extended to other radiation sources and tumor sites by following a similar training process.},
keywords = {Brachytherapy, Colon, Computer, Computer-Assisted, Deep Learning, Female, Humans, Iridium Radioisotopes, Male, Monte Carlo Method, Neural Networks, Organs at Risk, Prostate, Prostatic Neoplasms, Radiotherapy Dosage, Radiotherapy Planning, Rectum, Retrospective Studies, Sigmoid, Urinary Bladder, Uterine Cervical Neoplasms},
pubstate = {published},
tppubtype = {article}
}
METHODS AND MATERIALS: RapidBrachyDL, a 3-dimensional deep convolutional neural network (CNN) model, is proposed to predict dose distributions calculated with the MC method given a patient's computed tomography images, contours of clinical target volume (CTV) and organs at risk, and treatment plan. Sixty-one patients with prostate cancer and 10 patients with cervical cancer were included in this study, with data from 47 patients with prostate cancer being used to train the model.
RESULTS: Compared with ground truth MC simulations, the predicted dose distributions by RapidBrachyDL showed a consistent shape in the dose-volume histograms (DVHs); comparable DVH dosimetric indices including 0.73% difference for prostate CTV D90, 1.1% for rectum D2cc, 1.45% for urethra D0.1cc, and 1.05% for bladder D2cc; and substantially smaller prediction time, acceleration by a factor of 300. RapidBrachyDL also demonstrated good generalization to cervical data with 1.73%, 2.46%, 1.68%, and 1.74% difference for CTV D90, rectum D2cc, sigmoid D2cc, and bladder D2cc, respectively, which was unseen during the training.
CONCLUSION: Deep CNN-based dose estimation is a promising method for patient-specific brachytherapy dosimetry. Desired radiation quantities can be obtained with accuracies arbitrarily close to those of the source MC algorithm, but with much faster computation times. The idea behind deep CNN-based dose estimation can be safely extended to other radiation sources and tumor sites by following a similar training process.
2019
Shoemaker, Tristan; Vuong, Té; Glickman, Harry; Kaifi, Samar; Famulari, Gabriel; Enger, Shirin A.
Dosimetric Considerations for Ytterbium-169, Selenium-75, and Iridium-192 Radioisotopes in High-Dose-Rate Endorectal Brachytherapy Journal Article
In: International Journal of Radiation Oncology, Biology, Physics, vol. 105, no. 4, pp. 875–883, 2019, ISSN: 1879-355X.
Abstract | Links | BibTeX | Tags: Brachytherapy, Femur, Humans, Iridium Radioisotopes, Monte Carlo Method, Organs at Risk, Pelvic Bones, Radioisotopes, Radiotherapy Dosage, Rectal Neoplasms, Rectum, Selenium Radioisotopes, Tomography, Urinary Bladder, X-Ray Computed, Ytterbium
@article{shoemaker_dosimetric_2019,
title = {Dosimetric Considerations for Ytterbium-169, Selenium-75, and Iridium-192 Radioisotopes in High-Dose-Rate Endorectal Brachytherapy},
author = {Tristan Shoemaker and Té Vuong and Harry Glickman and Samar Kaifi and Gabriel Famulari and Shirin A. Enger},
doi = {10.1016/j.ijrobp.2019.07.003},
issn = {1879-355X},
year = {2019},
date = {2019-11-01},
journal = {International Journal of Radiation Oncology, Biology, Physics},
volume = {105},
number = {4},
pages = {875--883},
abstract = {PURPOSE: To investigate differences between prescribed and postimplant calculated dose in 192Ir high-dose-rate endorectal brachytherapy (HDR-EBT) by evaluating dose to clinical target volume (CTV) and organs at risk (OARs) calculated with a Monte Carlo-based dose calculation software, RapidBrachyMC. In addition, dose coverage, conformity, and homogeneity were compared among the radionuclides 192Ir, 75Se, and 169Yb for use in HDR-EBT.
METHODS AND MATERIALS: Postimplant dosimetry was evaluated using 23 computed tomography (CT) images from patients treated with HDR-EBT using the 192Ir microSelectron v2 (Elekta AB, Stockholm, Sweden) source and the Intracavitary Mold Applicator Set (Elekta AB, Stockholm, Sweden), which is a flexible applicator capable of fitting a tungsten rod for OAR shielding. Four tissue segmentation schemes were evaluated: (1) TG-43 formalism, (2) materials and nominal densities assigned to contours of foreign objects, (3) materials and nominal densities assigned to contoured organs in addition to foreign objects, and (4) materials specified as in (3) but with voxel mass densities derived from CT Hounsfield units. Clinical plans optimized for 192Ir were used, with the results for 75Se and 169Yb normalized to the D90 of the 192Ir clinical plan. RESULTS: In comparison to segmentation scheme 4, TG-43-based dosimetry overestimates CTV D90 by 6% (P = .00003), rectum D50 by 24% (P = .00003), and pelvic bone D50 by 5% (P = .00003) for 192Ir. For 169Yb, CTV D90 is overestimated by 17% (P = .00003) and rectum D50 by 39% (P = .00003), and pelvic bone D50 is significantly underestimated by 27% (P = .007). Postimplant dosimetry calculations also showed that a 169Yb source would give 20% (P = .00003) lower rectum V60 and 17% (P = .00008) lower rectum D50.
CONCLUSIONS: Ignoring high-Z materials in dose calculation contributes to inaccuracies that may lead to suboptimal dose optimization and disagreement between prescribed and calculated dose. This is especially important for low-energy radionuclides. Our results also show that with future magnetic resonance imaging-based treatment planning, loss of CT density data will only affect calculated dose in nonbone OARs by 2% or less and bone OARs by 13% or less across all sources if material composition and nominal mass densities are correctly assigned.},
keywords = {Brachytherapy, Femur, Humans, Iridium Radioisotopes, Monte Carlo Method, Organs at Risk, Pelvic Bones, Radioisotopes, Radiotherapy Dosage, Rectal Neoplasms, Rectum, Selenium Radioisotopes, Tomography, Urinary Bladder, X-Ray Computed, Ytterbium},
pubstate = {published},
tppubtype = {article}
}
METHODS AND MATERIALS: Postimplant dosimetry was evaluated using 23 computed tomography (CT) images from patients treated with HDR-EBT using the 192Ir microSelectron v2 (Elekta AB, Stockholm, Sweden) source and the Intracavitary Mold Applicator Set (Elekta AB, Stockholm, Sweden), which is a flexible applicator capable of fitting a tungsten rod for OAR shielding. Four tissue segmentation schemes were evaluated: (1) TG-43 formalism, (2) materials and nominal densities assigned to contours of foreign objects, (3) materials and nominal densities assigned to contoured organs in addition to foreign objects, and (4) materials specified as in (3) but with voxel mass densities derived from CT Hounsfield units. Clinical plans optimized for 192Ir were used, with the results for 75Se and 169Yb normalized to the D90 of the 192Ir clinical plan. RESULTS: In comparison to segmentation scheme 4, TG-43-based dosimetry overestimates CTV D90 by 6% (P = .00003), rectum D50 by 24% (P = .00003), and pelvic bone D50 by 5% (P = .00003) for 192Ir. For 169Yb, CTV D90 is overestimated by 17% (P = .00003) and rectum D50 by 39% (P = .00003), and pelvic bone D50 is significantly underestimated by 27% (P = .007). Postimplant dosimetry calculations also showed that a 169Yb source would give 20% (P = .00003) lower rectum V60 and 17% (P = .00008) lower rectum D50.
CONCLUSIONS: Ignoring high-Z materials in dose calculation contributes to inaccuracies that may lead to suboptimal dose optimization and disagreement between prescribed and calculated dose. This is especially important for low-energy radionuclides. Our results also show that with future magnetic resonance imaging-based treatment planning, loss of CT density data will only affect calculated dose in nonbone OARs by 2% or less and bone OARs by 13% or less across all sources if material composition and nominal mass densities are correctly assigned.
2014
Adams, Quentin E.; Xu, Jinghzu; Breitbach, Elizabeth K.; Li, Xing; Enger, Shirin A.; Rockey, William R.; Kim, Yusung; Wu, Xiaodong; Flynn, Ryan T.
Interstitial rotating shield brachytherapy for prostate cancer Journal Article
In: Medical Physics, vol. 41, no. 5, pp. 051703, 2014, ISSN: 2473-4209.
Abstract | Links | BibTeX | Tags: Brachytherapy, Catheters, Computer-Assisted, Equipment Design, Gadolinium, Humans, Iridium Radioisotopes, Male, Monte Carlo Method, Needles, Nickel, Platinum Compounds, Prostatic Neoplasms, Radiation Protection, Radioisotopes, Radiotherapy Dosage, Radiotherapy Planning, Rectum, Time Factors, Titanium, Urethra, Urinary Bladder
@article{adams_interstitial_2014,
title = {Interstitial rotating shield brachytherapy for prostate cancer},
author = {Quentin E. Adams and Jinghzu Xu and Elizabeth K. Breitbach and Xing Li and Shirin A. Enger and William R. Rockey and Yusung Kim and Xiaodong Wu and Ryan T. Flynn},
doi = {10.1118/1.4870441},
issn = {2473-4209},
year = {2014},
date = {2014-05-01},
journal = {Medical Physics},
volume = {41},
number = {5},
pages = {051703},
abstract = {PURPOSE: To present a novel needle, catheter, and radiation source system for interstitial rotating shield brachytherapy (I-RSBT) of the prostate. I-RSBT is a promising technique for reducing urethra, rectum, and bladder dose relative to conventional interstitial high-dose-rate brachytherapy (HDR-BT).
METHODS: A wire-mounted 62 GBq(153)Gd source is proposed with an encapsulated diameter of 0.59 mm, active diameter of 0.44 mm, and active length of 10 mm. A concept model I-RSBT needle/catheter pair was constructed using concentric 50 and 75 μm thick nickel-titanium alloy (nitinol) tubes. The needle is 16-gauge (1.651 mm) in outer diameter and the catheter contains a 535 μm thick platinum shield. I-RSBT and conventional HDR-BT treatment plans for a prostate cancer patient were generated based on Monte Carlo dose calculations. In order to minimize urethral dose, urethral dose gradient volumes within 0-5 mm of the urethra surface were allowed to receive doses less than the prescribed dose of 100%.
RESULTS: The platinum shield reduced the dose rate on the shielded side of the source at 1 cm off-axis to 6.4% of the dose rate on the unshielded side. For the case considered, for the same minimum dose to the hottest 98% of the clinical target volume (D(98%)), I-RSBT reduced urethral D(0.1cc) below that of conventional HDR-BT by 29%, 33%, 38%, and 44% for urethral dose gradient volumes within 0, 1, 3, and 5 mm of the urethra surface, respectively. Percentages are expressed relative to the prescription dose of 100%. For the case considered, for the same urethral dose gradient volumes, rectum D(1cc) was reduced by 7%, 6%, 6%, and 6%, respectively, and bladder D(1cc) was reduced by 4%, 5%, 5%, and 6%, respectively. Treatment time to deliver 20 Gy with I-RSBT was 154 min with ten 62 GBq (153)Gd sources.
CONCLUSIONS: For the case considered, the proposed(153)Gd-based I-RSBT system has the potential to lower the urethral dose relative to HDR-BT by 29%-44% if the clinician allows a urethral dose gradient volume of 0-5 mm around the urethra to receive a dose below the prescription. A multisource approach is necessary in order to deliver the proposed (153)Gd-based I-RSBT technique in reasonable treatment times.},
keywords = {Brachytherapy, Catheters, Computer-Assisted, Equipment Design, Gadolinium, Humans, Iridium Radioisotopes, Male, Monte Carlo Method, Needles, Nickel, Platinum Compounds, Prostatic Neoplasms, Radiation Protection, Radioisotopes, Radiotherapy Dosage, Radiotherapy Planning, Rectum, Time Factors, Titanium, Urethra, Urinary Bladder},
pubstate = {published},
tppubtype = {article}
}
METHODS: A wire-mounted 62 GBq(153)Gd source is proposed with an encapsulated diameter of 0.59 mm, active diameter of 0.44 mm, and active length of 10 mm. A concept model I-RSBT needle/catheter pair was constructed using concentric 50 and 75 μm thick nickel-titanium alloy (nitinol) tubes. The needle is 16-gauge (1.651 mm) in outer diameter and the catheter contains a 535 μm thick platinum shield. I-RSBT and conventional HDR-BT treatment plans for a prostate cancer patient were generated based on Monte Carlo dose calculations. In order to minimize urethral dose, urethral dose gradient volumes within 0-5 mm of the urethra surface were allowed to receive doses less than the prescribed dose of 100%.
RESULTS: The platinum shield reduced the dose rate on the shielded side of the source at 1 cm off-axis to 6.4% of the dose rate on the unshielded side. For the case considered, for the same minimum dose to the hottest 98% of the clinical target volume (D(98%)), I-RSBT reduced urethral D(0.1cc) below that of conventional HDR-BT by 29%, 33%, 38%, and 44% for urethral dose gradient volumes within 0, 1, 3, and 5 mm of the urethra surface, respectively. Percentages are expressed relative to the prescription dose of 100%. For the case considered, for the same urethral dose gradient volumes, rectum D(1cc) was reduced by 7%, 6%, 6%, and 6%, respectively, and bladder D(1cc) was reduced by 4%, 5%, 5%, and 6%, respectively. Treatment time to deliver 20 Gy with I-RSBT was 154 min with ten 62 GBq (153)Gd sources.
CONCLUSIONS: For the case considered, the proposed(153)Gd-based I-RSBT system has the potential to lower the urethral dose relative to HDR-BT by 29%-44% if the clinician allows a urethral dose gradient volume of 0-5 mm around the urethra to receive a dose below the prescription. A multisource approach is necessary in order to deliver the proposed (153)Gd-based I-RSBT technique in reasonable treatment times.