ABSTRACT
Objective:To establish a radiotherapy treatment planning process of high ventilation functional lung avoided (HVFLA) for thoracic tumors based on 4D-CT lung ventilation functional images and determine the treatment planning strategy of HVFLA radiotherapy, and so as to provide support for the clinical trials of HVFLA radiotherapy in thoracic cancer patients.Methods:A deep learning-based 4D-CT lung ventilation functional imaging model was established and integrated into the radiotherapy treatment planning process. Furthermore, ten thoracic cancer patients with 4D-CT simulation positioning were retrospectively enrolled in this study. The established model was used to obtain the 4D-CT lung ventilation functional imaging for each patient. According to the relative value of lung ventilation, the lung ventilation areas are equally segmented into high, medium and low lung ventilation and then imported them into Pinnacle 3 treatment planning system. According to the prescription dose of target and dose constraints of organ at risks (OARs), the clinical and HVFLA treatment plans were designed for each patient using volumetric modulated radiotherapy technique, and each plan should meet the clinical requirements and adding dose constraints of high ventilation functional lung for HVFLA plan. The dosimetric indexes of the target, OARs (lungs, heart and cord) and high functional lung (HFL) were used to evaluated the plan quality. The dosimetric indexes included D2, D98 and mean dose of target, V5, V10, V20, V30 and mean dose of lungs and HFL, V30, V40 and mean dose of heart, and D1 cm 3 of cord. Paired samples t-test was used for statistical analysis of the two groups of plans. Results:The target and OARs of the clinical plan and HVFLA plan meet the clinical requirements. The HVFLA plan resulted in a statistically significant reduction in the mean dose, V5, V10, V20, and V30 of the high functional lung by 1.2 Gy, 5.9%, 4.2%, 2.6%, and 2.3%, respectively ( t=-8.07, 4.02, -6.02, -7.06, -6.77, P<0.05). There was no statistical difference in the dosimetric indexes of lungs, heart and cord. Conclusions:We established the treatment planning process of HVFLA radiotherapy based on 4D-CT lung ventilation functional images. The HVFLA plan can effectively reduce the dose of HFL, while the doses of lungs, heart and cord had no significant difference compared with the clinical plan. The strategy of HVFLA radiotherapy planning is feasible to provide support for the implementation of HVFLA radiotherapy in thoracic cancer patients.
ABSTRACT
Objective:To verify the feasibility of using Elekta accelerated go live (AGL) standard process for the acceptance of multiple accelerators.Methods:The beams of three accelerators were adjusted by PTW Beamscan three-dimensional water tank to reach the AGL standard. Dose verification was performed for three accelerators that met AGL standards. A simple field test example from Cancer Hospital Chinese Academy of Medical Sciences was used to compare the MapCheck 3 surface dose measurement results with the surface dose calculated by the same accelerator model. Images of 10 patients including head and neck, esophagus, breast, lung and rectum were randomly selected. volumetric-modulated arc therapy (VMAT) and intensity modulated radiation therapy (IMRT) treatment techniques were used for planning design, and the measured dose of ArcCheck was compared with the planned dose calculated by the same accelerator model. One-way ANOVA was used to statistically analyze the passing rates of two-dimensional and three-dimensional dose verification.Results:The 6 MV X-ray percentage depth dose at 10 cm underwater (PDD 10) of three accelerators was 67.45%, 67.36%, 67.47%, and the maximum deviation between the three accelerators was 0.11%. The 6 MV flattenting filter free (FFF) mode X-ray PDD 10 was 67.33%, 67.20%, 67.20%, and the maximum deviation between the three accelerators was 0.13%. All required discrete point doses on each energy 30 cm×30 cm Profile spindle of the three accelerator X-rays deviated less than ±1% from the standard data. Absolute γ analysis was performed on the results of MapCheck 3 two-dimensional dose matrix validation. Under the 10% threshold of 2 mm/3% standard, the average passing rate of the test cases in Cancer Hospital Chinese Academy of Medical Sciences was above 99%, and the difference was not statistically significant ( P>0.05). Absolute γ analysis was performed on the ArcCheck verification results. Under the 10% threshold, the pass rate of 2 mm/3% was all above 95%, the maximum average passing rate of the three accelerators with different energy and different treatment techniques was 0.28% (6 MV, VMAT), 0.19%(6 MV FFF, VMAT), 0.56% (6 MV, IMRT) and 0.05% (6 MV FFF, IMRT), and the difference was not statistically significant ( P>0.05). Conclusion:Compared with traditional accelerator acceptance process, the acceptance time of each accelerator is shortened by 4-6 weeks by using the AGL standard process, and the radiotherapy plan of patients can be interchangeably executed among different accelerators.
ABSTRACT
Objective:To develop a deep learning method to predict the anatomical images of nasopharyngeal carcinoma patients during the treatment course, which could detect the anatomical variation for specific patients in advance.Methods:Imaging data including planning CT (pCT) and cone-beam CT (CBCT) for each fraction of 230 patients with T 3-T 4 staging nasopharyngeal carcinoma who treated in Cancer Hospital Chinese Academy of Medical Sciences from January 1, 2020 to December 31, 2022 were collected. The anatomical images of week k+1 were predicted using a 3D Unet model with inputs of pCT, CBCT on days 1-3, and CBCT of weeks 2- k. In this experiment, we trained four models to predict anatomical images of weeks 3-6, respectively. The nasopharynx gross tumor volume (GTV nx) and bilateral parotid glands were delineated on the predicted and real images (ground truth). The performance of models was evaluated by the consistence of the delineation between the predicted and ground truth images. Results:The proposed method could predict the anatomical images over the radiotherapy course. The contours of interest in the predicted image were consistent with those in the real image, with Dice similarity coefficient of 0.96, 0.90, 0.92, mean Hausdorff distance of 3.28, 4.18 and 3.86 mm, and mean distance to agreement of 0.37, 0.70, and 0.60 mm, for GTV nx, left parotid, and right parotid, respectively. Conclusion:This deep learning method is an accurate and feasible tool for predicting the patient's anatomical images, which contributes to predicting and preparing treatment strategy in advance and achieving individualized treatment.
ABSTRACT
In recent years, proton therapy technology has developed rapidly, and the number of patients treated with proton therapy has gradually increased. However, the application of proton therapy technology was far from practical needs. Because of the shortage of resources and the high cost, proton therapy systems are not accessible and affordable for most patients. In order to change this situation, it is necessary to develop a new truly practical proton therapy system based on clinical needs. Conceptual design of a practical proton therapy system was proposed. Compared with the existing system, one feature of the newly designed system is to reduce the maximum energy of the proton beam to 175~200 MeV; another feature is the configuration of deluxe and economical treatment rooms, the deluxe room is equipped with a rotating gantry and a six-dimensional treatment bed, and the economical room is equipped with a horizontal fixed beam and a patient vertical rotating setup device. This design can not only reduce the cost of proton therapy system and equipment room construction, but also facilitate the hospital to choose the appropriate configuration, which will ultimately benefit more patients.
Subject(s)
Humans , Proton Therapy , Radiotherapy Planning, Computer-Assisted , Hospitals , Radiotherapy DosageABSTRACT
Objective:To investigate a time series deep learning model for respiratory motion prediction.Methods:Eighty pieces of respiratory motion data from lung cancer patients were used in this study. They were divided into a training set and a test set at a ratio of 8∶2. The Informer deep learning network was employed to predict the respiratory motions with a latency of about 600 ms. The model performance was evaluated based on normalized root mean square errors (nRMSEs) and relative root mean square errors (rRMSEs).Results:The Informer model outperformed the conventional multilayer perceptron (MLP) and long short-term memory (LSTM) models. The Informer model yielded an average nRMSE and rRMSE of 0.270 and 0.365, respectively, at a prediction time of 423 ms, and 0.380 and 0.379, respectively, at a prediction time of 615 ms.Conclusions:The Informer model performs well in the case of a longer prediction time and has potential application value for improving the effects of the real-time tracking technology.
ABSTRACT
Compared with conventional radiotherapy, FLASH radiotherapy has advantages in protecting normal tissues, while the dose rate is increased by more than 100 times. If the shielding design of the treatment room is carried out according to the existing standard, the thickness and cost of the shielding wall will be significantly increased, or even hardly to meet the requirement of the standards, resultsing in the failure of the application of FLASH radiotherapy. By investigating the domestic and foreign standards and literature, this paper analyzes the challenges brought by FLASH radiotherapy technology to the shielding design of radiotherapy treatment room in China. Dose rate control standards adopted by different countries in the shielding design are emphatically compared as well. In several countries, the average dose rate under the actual treatment conditions was considered in the shielding design. In China, the method of instantaneous dose rate taking acount of occupancy factor is adopted. However, if FLASH radiotherapy technology is applied, the requirement of instantaneous dose rate will be difficult to meet. In order to improve the high dose rate radiotherapy technology such as FLASH radiotherapy, the revision of the existing standards is advised if the authorized limits are not changed. To use the average dose rate limit within a certain period of time for control, or to raise the control standard in the case of flash radiotherapy, are also avaliable.
ABSTRACT
Objective:To investigate the pseudo-CT generation from cone beam CT (CBCT) by a deep learning method for the clinical need of adaptive radiotherapy.Methods:CBCT data from 74 prostate cancer patients collected by Varian On-Board Imager and their simulated positioning CT images were used for this study. The deformable registration was implemented by MIM software. And the data were randomly divided into the training set ( n=59) and test set ( n=15). U-net, Pix2PixGAN and CycleGAN were employed to learn the mapping from CBCT to simulated positioning CT. The evaluation indexes included mean absolute error (MAE), structural similarity index (SSIM) and peak signal to noise ratio (PSNR), with the deformed CT chosen as the reference. In addition, the quality of image was analyzed separately, including soft tissue resolution, image noise and artifacts, etc. Results:The MAE of images generated by U-net, Pix2PixGAN and CycleGAN were (29.4±16.1) HU, (37.1±14.4) HU and (34.3±17.3) HU, respectively. In terms of image quality, the images generated by U-net and Pix2PixGAN had excessive blur, resulting in image distortion; while the images generated by CycleGAN retained the CBCT image structure and improved the image quality.Conclusion:CycleGAN is able to effectively improve the quality of CBCT images, and has potential to be used in adaptive radiotherapy.
ABSTRACT
In recent years, the issue of "reproducibility" of scientific research results has become more and more prominent. Radiobiology is a medical science that studies the biological effect of radiation on living organisms, and there is also a serious problem of "reproducibility of findings". Inaccuracy of physical dose or incomplete dosimetric reports is one of the main causes. Use of guidelines, specifications and recommendations for dosimetric measurement, such as the standardized scoring system for dosimetric reports, will help improving the standardization and accuracy of physical dose measurement in radiobiological research. In this article, multiple guidelines and recommends for improving collaboration between radiobiology and radiation physics, as well as for dose standardization of radiobiological research were evaluated, aiming to provide reference for improving the reproducibility of radiobiological research.
ABSTRACT
Objective:To observe the effect of project-based learning (PBL) in the clinical teaching of radiation physics.Methods:Thirty-two residents specializing in radiotherapy were included in the study. In the experimental group ( n=16), PBL was adopted, while traditional clinical teaching method was employed in the control group ( n=16). After the rotation, the assessment was conducted, as well as a questionnaire survey was performed, including five aspects: overall satisfaction, understanding of radiation physics knowledge, learning motivation, learning burden, and learning efficiency. Results:The assessment score in the experimental group was 86.31±5.41, which was higher than 75.28±5.91 in the control group, and the difference was statistically significant. Residents in the experimental group were satisfied with the effect of PBL.Conclusion:Compared with the traditional teaching method, PBL can improve the learning motivation, efficiency, and performance of radiotherapy residents, which is highly recognized by the residents.
ABSTRACT
Objective:To introduce the clinical dosimetry commissioning methods and results of the 1.5 T MR-linac.Methods:In May, 2019, an Elekta Unity 1.5 T MR-linac was installed in Cancer Hospital, Chinese Academy of Medical Sciences and dosimetry commissioning was performed with magnetic field compatible measuring instruments. Commissioning items include absolute dose calibration, data acquisition and planning system model verification.Results:Absolute dose calibration in magnetic field should be corrected by magnetic field correction factor. The standard output dose of Unity was 87 cGy. Gamma analysis (3%/2 mm) was performed on the beam collection data and the planning system calculation data. The average pass rate of dose verification of standard field test cases was 96.41%, and the TG119 test case was 98.24%. The IROC end to end test case was 97.5%(7%/4 mm).Conclusions:The planning system model and the beam collection data have good consistency. The dose verification results of the standard field and TG119 test cases meet the general tolerance limit requirements of the AAPM TG218 report, and the verification results of the IROC end-to-end test cases meet the IROC center standards.
ABSTRACT
Non-coplanar radiotherapy is a kind of radiotherapy technology which employs multiple non-coplanar fixed fields or non-coplanar arcs. The non-coplanar field can be defined that the central axis of each field is not on the same plane, while the non-coplanar arc can be described that the trajectory formed by each arc is not on the same plane. Compared with coplanar radiotherapy, non-coplanar radiotherapy can achieve multi-angle or multi-radian irradiation, which effectively improves the focusing level of ray and is beneficial to enlarge the radiation dose of the target area between the surrounding normal tissues. Its dosimetric advantages have been proven in multiple types of tumors, such as intracranial tumors, liver cancer and lung cancer, etc. Multiple approaches can be employed to realize non-coplanar radiotherapy, which can be divided into the non-coplanar conic radiotherapy, non-coplanar conformal radiotherapy, non-coplanar intensity-modulated radiotherapy and non-coplanar volumetric modulated arc therapy according to the established sequence. In this review, the development process and principal characteristics of these implementations were summarized.
ABSTRACT
Powered by big data and artificial intelligence, the research and clinical application of treatment planning automation for radiation therapy are rapidly growing. The application and supervision of planning automation systems necessitate careful consideration of different levels of automation, as well as the context for use. For autonomous vehicles, the levels of automation have been defined at home and abroad. Nevertheless, no such definitions exist for radiotherapy planning automation. To promote and standardize the development of radiotherapy planning automation and initiate discussion within the community, we developed this recommendation with reference to the taxonomy of driving automation for vehicles and divided the radiotherapy planning automation into six levels (level 1 to 6).
ABSTRACT
PURPOSE: Radiation pneumonitis (RP) is the main source of toxicity in thoracic radiotherapy. This study proposed a deep learning-based dual-omics model, which aims to improve the RP prediction performance by integrating more data points and exploring the data in greater depth. MATERIALS AND METHODS: The bimodality data were the original dose (OD) distribution and the ventilation image (VI) derived from four-dimensional computed tomography (4DCT). The functional dose (FD) distribution was obtained by weighting OD with VI. A pre-trained three-dimensional convolution (C3D) network was used to extract the features from FD, VI, and OD. The extracted features were then filtered and selected using entropy-based methods. The prediction models were constructed with four most commonly used binary classifiers. Cross-validation, bootstrap, and nested sampling methods were adopted in the process of training and hyper-tuning. RESULTS: Data from 217 thoracic cancer patients treated with radiotherapy were used to train and validate the prediction model. The 4DCT-based VI showed the inhomogeneous pulmonary function of the lungs. More than half of the extracted features were singular (of none-zero value for few patients), which were eliminated to improve the stability of the model. The area under curve (AUC) of the dual-omics model was 0.874 (95% confidence interval: 0.871-0.877), and the AUC of the single-omics model was 0.780 (0.775-0.785, VI) and 0.810 (0.804-0.811, OD), respectively. CONCLUSIONS: The dual-omics outperformed single-omics for RP prediction, which can be contributed to: (1) using more data points; (2) exploring the data in greater depth; and (3) incorporating of the bimodality data.
Subject(s)
Carcinoma, Non-Small-Cell Lung , Deep Learning , Lung Neoplasms , Radiation Pneumonitis , Four-Dimensional Computed Tomography , Humans , Lung Neoplasms/diagnostic imaging , Lung Neoplasms/radiotherapy , Radiation Pneumonitis/etiologyABSTRACT
Objective:To evaluate the application of visual feedback coaching method, which is embedded in an optical surface monitoring system, in deep inspiration breath holding during the radiotherapy in left breast cancer patients after breast-conserving surgery.Methods:Thirty patients with left breast cancer, who were scheduled to receive the whole breast radiotherapy after breast-conserving surgery, met the requirements of deep inspiration breath holding after respiratory coaching with the visual feedback coaching module in the optical surface monitoring system. Active breathing control equipment was used to control breath-holding state and CT simulation was performed. During treatment, optical surface monitoring system was used to guide radiotherapy. All patients were randomly divided into two groups. In group A ( n=15), visual feedback respiratory training method was utilized and not employed in group B ( n=15). In group A, the visual feedback coaching bar of the optical surface monitoring system was implemented, while audio interactive method was employed to guide patients to hold their breath. Real-time data of optical body surface monitoring were used to compare the interfraction reproducibility and intrafraction stability of breath holding fraction between two groups. Besides, the number of breath holding and treatment time per fraction were also compared. GraphPad prism 6.0 software was used for data processing and mapping, and SPSS 21.0 software was used for analyzing mean value and normality testing. Results:Compared with the control group, the reproducibility in the experiment group was reduced from 1.5 mm to 0.7 mm, the stability was reduced from 1.1 mm to 0.8 mm, the mean number of breath holding required per fraction was decreased from 4.6 to 2.4, the mean beam-on time per fraction from 336 s to 235 s, and the treatment time per fraction was shortened from 847 s to 602 s (all P<0.05), respectively. Conclusions:The application of visual feedback coaching method can improve the reproducibility and stability of breath holding during radiotherapy for left breast cancer, and it can also effectively reduce the number of breath holding and shorten the treatment time per fraction.
ABSTRACT
Objective:To design a drum-shaped applicator through Monte Carlo simulation for breast intraoperative radiotherapy.Methods:Applicator designing process is as follows: first, determined the shape of the applicator based on the geometric characteristics of the breast tumor bed closed to the chest wall; second, calculated the scattering angle and dose rate of the electron beam after passing through a series of scattering foils of different thicknesses to determine the thickness of the scattering foil; thrid, modeled the layer according to the applicator′s geometric characteristics where modulator located, and designed the modulator through the relationship between the geometric characteristics of the layer and the surface dose of the applicator. EGSnrc/BEAMnrc and EGS4/DOSXYZ were employed to model the head of the Mobetron, the layer, the applicator, and to calculate the dose distributions.Results:The applicator has two components. The upper component is a 3cm-diametre cylindrical collimator with 0.5cm wall made of 0.3cm steel and 0.2cm water equivalent material (WEM), a 0.13cm-foil made of tansgen. The lower component is a 4cm-diametre drum made of 0.2cm WEM and a 0.14cm maximum thickness hill-shaped modulator made of steel. When the energy of electron beam was 12MeV, the dose rate was about 90.44 cGy/min, and the depth of the 50% isodose curve was 1cm.Conclusion:The applicator is successfully designed, and can obtain a drum-shaped dose distribution.
ABSTRACT
By using optical surface guided radiotherapy technology, the principle of three-dimensional body surface imaging is employed to obtain body surface images in a real-time manner. By comparing with reference images, it can verify the position before treatment, and realize real-time monitoring and gated treatment during treatment. It is a non-invasive and non-radiation technology, which is mainly applied in the treatment of intracranial, head and neck, chest and abdomen, breast, extremities and pediatric tumors. The research progresses consist of four aspects including less body surface markers, less restraint fixation, safer collision prediction and more accurate real-time tracking.
ABSTRACT
Objective:To establish an automatic planning method using volumetric-modulated arc therapy (VMAT) for primary liver cancer (PLC) radiotherapy based on predicting the feasibility dose-volume histogram (DVH) and evaluate its performance.Methods:Ten patients with PLC were randomly chosen in this retrospective study. Pinnacle Auto-Planning was used to design the VMAT automatic plan, and the feasibility DVH curve was obtained through the PlanIQ dose prediction, and the initial optimization objectives of the automatic plan were set according to the displayed feasible objectives interval. The plans were accessed according to dosimetric parameters of the planning target volume and organs at risk as well as the monitor units. All patients′ automatic plans were compared with clinically accepted manual plans by using the paired t-test. Results:There was no significant difference of the planning target volume D 2%, D 98%, D mean or homogeneity index between the automatic and manual plans ((58.55±2.81) Gy vs.(57.98±4.17) Gy, (47.15±1.58) Gy vs.(47.82±1.38) Gy, (53.14±0.95) Gy vs.(53.44±1.67) Gy and 1.15±0.05 vs. 1.14±0.07, all P>0.05). The planning target volume conformity index of the manual plan was slightly higher than that of the automatic plan (0.77±0.08 vs. 0.69±0.06, P<0.05). The mean doses of normal liver, V 30Gy, V 20Gy, V 10Gy, V 5Gy and V< 5Gy of the automatic plan were significantly better than those of the manual plan ((26.68±11.13)% vs.(28.00±10.95)%, (29.96±11.50)% vs.(31.89±11.51)%, (34.88±11.51)% vs.(38.66±11.67)%, (45.38±12.40)% vs.(50.74±13.56)%, and (628.52±191.80) cm 3vs.(563.15±188.39) cm 3, all P<0.05). The mean doses of the small intestine, the duodenum, and the heart, as well as lung V 10 of the automatic plan were significantly less than those of the manual plan ((1.83±2.17) Gy vs.(2.37±2.81) Gy, (9.15±9.36) Gy vs.(11.18±10.49) Gy, and (5.44±3.10) Gy vs.(6.25±3.26) Gy, as well as (12.70±7.08)% vs.(14.47±8.11)%, all P<0.05). Monitor units did not significantly differ between two plans ((710.67±163.72) MU vs.(707.53±155.89) MU, P>0.05). Conclusions:The automatic planning method using VMAT for PLC radiotherapy based on predicting the feasibility DVH enhances the quality for PLC plans, especially in terms of normal liver sparing. Besides, it also has advantages for the protection of the intestine, whole lung and heart.
ABSTRACT
Objective:To investigate the setup errors of postoperative radiotherapy immobilized with integrated cervicothoracic board (mask) system in breast cancer patients.Methods:Thirty-two breast cancer patients treated with postoperative radiotherapy immobilized with integrated cervicothoracic board (mask) system were prospectively recruited in this study. Breast/chest wall (cw) and supra/infraclavicular nodal region (sc) were irradiated with intensity-modulated radiotherapy. CBCT location verification in radiotherapy and target areas of the breast/chest wall and upper and lower collarbone were carried out, respectively. The consistency between setup errors and the position of the upper and lower target areas of 239 CBCT images was analyzed.Results:The translational setup errors of the breast/chest wall in the X-cw (left-right), Y-cw (superior-inferior), Z-cw (anterior-posterior) directions were (1.84±2.36) mm, (1.99±2.48) mm, and (1.75±1.86) mm, respectively. The translational setup errors of the supra/infraclavicular nodal region in the X-sc (left-right), Y-sc (superior-inferior), Z-sc (anterior-posterior) directions were (1.98±2.44) mm, (1.98±2.48) mm, and (1.71±1.79) mm, respectively. The differences of translational setup errors between the breast/chest wall and supra/infraclavicular nodal region in the X, Y, Z directions were (0.38±0.66) mm, (0.07±0.41) mm, and (0.45±0.92) mm, respectively. Conclusion:For the breast cancer patients treated with postoperative radiotherapy covering breast/chest wall and supra/infraclavicular nodal region, the integrated cervicothoracic board (mask) immobilization system provides good reproducibility and yields Sfew setup errors.
ABSTRACT
Objective:To analyze and compare the dosimetric differences based on volumetric-modulated arc therapy (VMAT), fixed field intensity-modulated radiotherapy (F_IMRT), and electron irradiation combined with VMAT (E&VMAT) in radiotherapy for breast cancer after modified mastectomy, aiming to provide reference for clinical selection of treatment plan.Methods:Ten patients with the left breast cancer who received radiotherapy after modified mastectomy were randomly selected. The target areas included chest wall and supraclavicular region, and the prescribed dose was 43.5 Gy in 15 fractions (2.9 Gy/F). Based on the Pinnacle 3 planning system, the VMAT, F_IMRT and E&VMAT plans (electron beam for chest wall, VMAT for supraclavicular area) were designed for each patient. The conformity and homogeneity of the target areas, the dose of organs at risk and treatment time were compared. Results:The VMAT plan could improve the dose distribution of the target areas. The conformity index and homogeneity index of the target dose were significantly better than those of the F_IMRT and E&VMAT plans (all P<0.05). The average dose, V 30Gy, V 20Gy, V 10Gy of the left lung in the VMAT plan were significantly better than those in the F_IMRT and E&VMAT plans (all P<0.05). The V 5Gy of the left lung in the VMAT plan was significantly better than that in the F_IMRT plan ( P<0.05). There was no statistical difference in the V 5Gy of the left lung between the VMAT and E&VMAT plans ( P>0.05). The heart, right breast and right lung of the VMAT plan could meet the clinical dose limit requirements. The treatment time of the VMAT, F_IMRT and E&VMAT plans was (326±27) s, (1 082±169) s, and (562±48) s, respectively. Conclusions:Compared with the F_IMRT and E&VMAT plans, the VMAT plan has better quality and shorter treatment time. VMAT plan has higher value in clinical application compared with the F_IMRT and E&VMAT plans.
ABSTRACT
PURPOSE: Convolutional neural networks (CNNs) offer a promising approach to automated segmentation. However, labeling contours on a large scale is laborious. Here we propose a method to improve segmentation continually with less labeling effort. METHODS: The cohort included 600 patients with nasopharyngeal carcinoma. The proposed method was comprised of four steps. First, an initial CNN model was trained from scratch to perform segmentation of the clinical target volume. Second, a binary classifier was trained using a secondary CNN to identify samples for which the initial model gave a dice similarity coefficient (DSC) < 0.85. Third, the classifier was used to select such samples from the new coming data. Forth, the final model was fine-tuned from the initial model, using only selected samples. RESULTS: The classifier can detect poor segmentation of the model with an accuracy of 92%. The proposed segmentation method improved the DSC from 0.82 to 0.86 while reducing the labeling effort by 45%. CONCLUSIONS: The proposed method reduces the amount of labeled training data and improves segmentation by continually acquiring, fine-tuning, and transferring knowledge over long time spans.