Three-dimensional dose prediction based on deep convolutional neural networks for brain cancer in CyberKnife: accurate beam modelling of homogeneous tissue.

Objectives: Accurate beam modelling is essential for dose calculation in stereotactic radiation therapy (SRT), such as CyberKnife treatment. However, the present deep learning methods only involve patient anatomical images and delineated masks for training. These studies generally focus on traditional intensity-modulated radiation therapy (RT) plans. Nevertheless, this paper aims to develop a deep CNN-based method for CyberKnife plan dose prediction about brain cancer patients. It utilized modelled beam information, target delineation, and patient anatomical information. Methods: This study proposes a method that adds beam information to predict the dose distribution of CyberKnife in brain cases. A retrospective dataset of 88 brain and abdominal cancer patients treated with the Ray-tracing algorithm was performed. The datasets include patients' anatomical information (planning CT), binary masks for organs at risk (OARs) and targets, and clinical plans (containing beam information). The datasets were randomly split into 68, 6, and 14 brain cases for training, validation, and testing, respectively. Results: Our proposed method performs well in SRT dose prediction. First, for the gamma passing rates in brain cancer cases, with the 2 mm/2% criteria, we got 96.7% ± 2.9% for the body, 98.3% ± 3.0% for the planning target volume, and 100.0% ± 0.0% for the OARs with small volumes referring to the clinical plan dose. Secondly, the model predictions matched the clinical plan's dose-volume histograms reasonably well for those cases. The differences in key metrics at the target area were generally below 1.0 Gy (approximately a 3% difference relative to the prescription dose). Conclusions: The preliminary results for selected 14 brain cancer cases suggest that accurate 3-dimensional dose prediction for brain cancer in CyberKnife can be accomplished based on accurate beam modelling for homogeneous tumour tissue. More patients and other cancer sites are needed in a further study to validate the proposed method fully. Advances in knowledge: With accurate beam modelling, the deep learning model can quickly generate the dose distribution for CyberKnife cases. This method accelerates the RT planning process, significantly improves its operational efficiency, and optimizes it.

Application of MR images in radiotherapy planning for brain tumor based on deep learning.

PURPOSE: Explore the function and dose calculation accuracy of MRI images in radiotherapy planning through deep learning methods. METHODS: 131 brain tumor patients undergoing radiotherapy with previous MR and CT images were recruited for this study. A new series of MRI from the aligned MR was firstly registered to CT images strictly using MIM software and then resampled. A deep learning method (U-NET) was used to establish a MRI-to-CT conversion model, for which 105 patient images were used as the training set and 26 patient images were used as the tuning set. Data from additional 8 patients were collected as the test set, and the accuracy of the model was evaluated from a dosimetric standpoint. RESULTS: Comparing the synthetic CT images with the original CT images, the difference in dosimetric parameters D98, D95, D2 and Dmean of PTV in 8 patients was less than 0.5%. The gamma passed rates of PTV and whole body volume were: 1%/1 mm: 93.96%±6.75%, 2%/2 mm: 99.87%±0.30%, 3%/3 mm: 100.00%±0.00%; and 1%/1 mm: 99.14%±0.80%, 2%/2 mm: 99.92%±0.08%, 3%/3 mm: 99.99%±0.01%. CONCLUSION: MR images can be used both in delineation and treatment efficacy evaluation and in dose calculation. Using the deep learning way to convert MR image to CT image is a viable method and can be further used in dose calculation.

[Study on Automatic Plan Method for Radiotherapy after Breast-conserving Surgery Based on TiGRT System].

To study an automatic plan(AP) method for radiotherapy after breast-conserving surgery based on TiGRT system and and compare with manual plan (MP). The dosimetry parameters of 10 patients and the evaluation of scoring table were analyzed, it was found that the targets dose of AP were better than that of MP, but there was no statistical difference except for CI, The V5, V20 and V30 of affected lungs and whole lungs in AP were lower than all that in MP, the Dmean of hearts was slightly higher than that of MP, but the difference was not statistically significant, the MU of AP was increase by 16.1% compared with MP, the score of AP evaluation was increase by 6.1% compared with MP. So the AP could be programmed and automated while ensuring the quality of the plan, and can be used to design the plans for radiotherapy after breast-conserving surgery.

A Clinical Study Investigating Whether the Tongue-Out Position Improves the Quality of the Anatomical Appearance of the Pharynx on CT Imaging.

Objective: To evaluate the effect of using the tongue-out position on the quality of the anatomical appearance of the pharynx on computed tomography (CT) images. Methods: The data from enhanced CT thin-section images of the head and neck in 119 cases scanned were retrospectively analyzed. The cases were divided into two groups based on the position of the tip of the tongue on the images: the tongue-out group (63 cases) and non-tongue-out group (56 cases). Two observers separately evaluated the anatomy of the soft palate, uvula, palatine tonsils, epiglottis, epiglottic fossa, pyriform fossa, arytenoid folds, and tongue on all images. The Kappa test was applied to assess the consistency of scores between the two observers. In the case of data that satisfied the normal distribution, the significance of the difference in the average scores between the two groups was tested using an independent samples t-test with a value of p > 0.05. In the case of data that did not satisfy the normal distribution, the Mann-Whitney U test was adopted to test the significance of the difference in the average scores between the two groups using a value of p < 0.05. The number of cases with swallowing artifacts on the CT images in both groups was statistically analyzed and the chi-square test was used to determine whether the difference in the incidence of artifacts between the two groups was significant. Results: The Kappa test showed good consistency between the two observers scoring of the soft palate, uvula, epiglottis, epiglottic fossa, pyriform fossa, aryepiglottic folds, and tongue. The image scores of the soft palate, uvula, epiglottis, epiglottic fossa, and tongue in the tongue-out group vs. the non-tongue-out group did not satisfy the normal distribution. The Mann-Whitney U test showed that the differences in the image scores between the two groups were statistically significant in all cases (p < 0.05). The incidence of swallowing artifacts in the tongue-out group and the non-tongue-out group was 15 and 32%, respectively. The result of the chi-square test showed that the difference in the incidence of swallowing artifacts between the two groups was statistically significant (p = 0.037). Conclusion: The tongue-out position facilitated an improvement in the CT appearance of pharyngeal anatomy and was associated with a reduction in the incidence of swallowing artifacts.

A Risk Prediction Model by LASSO for Radiation-Induced Xerostomia in Patients With Nasopharyngeal Carcinoma Treated With Comprehensive Salivary Gland-Sparing Helical Tomotherapy Technique.

OBJECTIVE: This study aimed to develop a least absolute shrinkage and selection operator (LASSO)-based multivariable normal tissue complication probability (NTCP) model to predict radiation-induced xerostomia in patients with nasopharyngeal carcinoma (NPC) treated with comprehensive salivary gland-sparing helical tomotherapy technique. METHODS AND MATERIALS: LASSO with the extended bootstrapping technique was used to build multivariable NTCP models to predict factors of patient-reported xerostomia relieved by 50% and 80% compared with the level at the end of radiation therapy within 1 year and 2 years, R50-1year and R80-2years, in 203 patients with NPC. The model assessment was based on 10-fold cross-validation and the area under the receiver operating characteristic curve (AUC). RESULTS: The prediction model by LASSO with 10-fold cross-validation showed that radiation-induced xerostomia recovery could be predicted by prognostic factors of R50-1year (age, gender, T stage, UICC/AJCC stage, parotid Dmean, oral cavity Dmean, and treatment options) and R80-2years (age, gender, T stage, UICC/AJCC stage, oral cavity Dmean, N stage, and treatment options). These prediction models also demonstrated a good performance by the AUC. CONCLUSION: The prediction models of R50-1year and R80-2years by LASSO with 10-fold cross-validation were recommended to validate the NTCP model before comprehensive salivary gland-sparing radiation therapy in patients with NPC.

Automatic segmentation of pelvic organs-at-risk using a fusion network model based on limited training samples.

Background: Efficient and accurate methods are needed to automatically segmenting organs-at-risk (OAR) to accelerate the radiotherapy workflow and decrease the treatment wait time. We developed and evaluated the use of a fused model Dense V-Network for its ability to accurately segment pelvic OAR.Material and methods: We combined two network models, Dense Net and V-Net, to establish the Dense V-Network algorithm. For the training model, we adopted 100 kV computed tomography (CT) images of patients with cervical cancer, including 80 randomly selected as training sets, by which to adjust parameters of the automatic segmentation model, and the remaining 20 as test sets to evaluate the performance of the convolutional neural network model. Three representative parameters were used to evaluate the segmentation results quantitatively.Results: Clinical results revealed that Dice similarity coefficient values of the bladder, small intestine, rectum, femoral head and spinal cord were all above 0.87 mm; and Jaccard distance was within 2.3 mm. Except for the small intestine, the Hausdorff distance of other organs was less than 9.0 mm. Comparison of our approaches with those of the Atlas and other studies demonstrated that the Dense V-Network had more accurate and efficient performance and faster speed.Conclusions: The Dense V-Network algorithm can be used to automatically segment pelvic OARs accurately and efficiently, while shortening patients' waiting time and accelerating radiotherapy workflow.

[A fusion network model based on limited training samples for the automatic segmentation of pelvic endangered organs].

When applying deep learning to the automatic segmentation of organs at risk in medical images, we combine two network models of Dense Net and V-Net to develop a Dense V-network for automatic segmentation of three-dimensional computed tomography (CT) images, in order to solve the problems of degradation and gradient disappearance of three-dimensional convolutional neural networks optimization as training samples are insufficient. This algorithm is applied to the delineation of pelvic endangered organs and we take three representative evaluation parameters to quantitatively evaluate the segmentation effect. The clinical result showed that the Dice similarity coefficient values of the bladder, small intestine, rectum, femoral head and spinal cord were all above 0.87 (average was 0.9); Jaccard distance of these were within 2.3 (average was 0.18). Except for the small intestine, the Hausdorff distance of other organs were less than 0.9 cm (average was 0.62 cm). The Dense V-Network has been proven to achieve the accurate segmentation of pelvic endangered organs.

Dose fall-off during the treatment of thoracic spine metastasis with CyberKnife stereotactic body radiation therapy (SBRT).

CyberKnife stereotactic body radiation therapy (SBRT) is becoming increasingly used for cancer treatment and, to maximize its clinical application, it is important to define the dosimetric characteristics, optimal dose, and fractionation regimens. The aim of this study was to evaluate the dose fall-off in two fractionated regimens of CyberKnife SBRT during the treatment of thoracic spinal metastasis. Patients with spinal metastasis involving a vertebra and pedicle were treated with 40 Gy in 5 fractions (n = 4), and patients with spinal metastasis involving only a vertebra received 33 Gy in 3 fractions (n = 4). A new approach was used to measure absolute dose fall-off distance, relative dose fall-off distance, and the dose fall-off per unit distance along four reference directions in the axial plane. Patients treated with 33 Gy/3 fractions had a greater absolute dose fall-off distance in direction 1 (from the point with maximum dose [Dmax] towards the spinal cord) and direction 3 (the opposite of direction 1), a greater relative dose fall-off distance in direction 3, and a lower dose fall-off per unit distance in direction 1 and 3 compared to patients treated with 40 Gy/5 fractions (all p < 0.05). Overall, the dose fall-off towards the spinal cord is rapid during the treatment of thoracic spinal metastasis with CyberKnife SBRT, which allows a higher dose of radiation to be delivered to the tumor and, at the same time, better protection of the spinal cord.

Reducing Xerostomia by Comprehensive Protection of Salivary Glands in Intensity-Modulated Radiation Therapy with Helical Tomotherapy Technique for Head-and-Neck Cancer Patients: A Prospective Observational Study.

OBJECTIVE: This study aimed to analyze the effects of comprehensive protection of bilateral parotid glands (PG-T), contralateral submandibular gland (cSMG), and accessory salivary glands in the oral cavity (OC) by helical tomotherapy for head-and-neck cancer patients. METHODS: Totally 175 patients with histologically confirmed head-and-neck cancer treated with helical tomotherapy were recruited. The doses delivered to PG-T, cSMG, and OC were constrained to be as low as possible in treatment planning. The saliva flow rates and xerostomia questionnaire were evaluated. Correlation between xerostomia and other clinical factors were assessed using univariate and multivariate models. The impact of salivary gland dose on locoregional (LR) recurrence was assessed by Cox analysis. ROC curve was used to determine the threshold of mean dose for each gland. RESULTS: The median follow-up was 25 (19-36) months. The OC mean dose, PG-T mean dose, cSMG mean dose, age, clinical stage (II and III versus IV), and both unstimulated and stimulated saliva flow rates were significantly correlated with xerostomia. The OC mean dose, cSMG mean dose, age, and clinical stage were predictors of xerostomia after adjusting PG-T mean dose, and unstimulated and stimulated saliva flow rates. Xerostomia was significantly decreased when the mean doses of PG-T, cSMG, and OC were kept below 29.12Gy, 29.29Gy, and 31.44Gy, respectively. At 18 months after radiation therapy, early LR recurrence rate was only 4%. CONCLUSION: Comprehensive protection of salivary glands minimized xerostomia in head-and-neck cancer patients treated by helical tomotherapy, without increasing early LR recurrence risk.

[Accuracy of Dose Calculation with CBCT Images under Different Scanning Conditions].

The treatment plans were designed based on planning CT of Catphan phantom's homogenic and heterogenetical module, respectively. With the OBI system, Catphan phantom was scanned under different scanning conditions. The dose was recalculated by applying treatment plans based on planning CT to the CBCT images with its individual hounsfield unit-electron density calibration curve. The dose distributions were compared with those of the original plan, the results of HI showed a good agreement. CIs in the homogenic module were superior to those in the heterogenetical one. There is a good dose distribution for CBCT images under different scanning conditions. It is helpful for CBCT images directly used for dose re-calculation in adaptive radiation therapy (ART).

[RapidArc Delivery Performance Using EDR2 Film and EPID Measurement].

This study describes the development and implementation of EDR2 film and electronic portal imaging device (EPID) during RapidArc QA. The tests were designed to evaluate RapidArc performance using EDR2 film and EPID tools. The accuracy of MLC position during gantry rotation, the ability to vary and control the dose-rate and gantry speed, the synchronization of variable MLC speed and dose-rate were examined. The picket fence test of MLC in stationary gantry and RapidArc modes were implemented. The film and EPID showed a good consistency. During the evaluation of MLC speed, gantry speed and dose-rates, the dose of different parts in a field showed a good agreement, with the mean deviation of 0.24%vs 0.19%. The analysis of dose value was less than 2%. This study demonstrated that EDR2 film and EPID system can be used as reliable and efficient quality assurance tools for RapidArc delivery performance. Of course, the use of VMAT QA with EPID increases the efficiency of routine QA.

[The Dose Effect of Isocenter Selection during IMRT Dose Verification with the 2D Chamber Array].

To investigate the dose effect of isocenter difference during IMRT dose verification with the 2D chamber array. The samples collected from 10 patients were respectively designed for IMRT plans, the isocenter of which was independently defined as P(o), P(x) and P(y). P(o) was fixed on the target center and the other points shifted 8cm from the target center in the orientation of x/y. The PTW729 was used for 2D dose verification in the 3 groups which beams of plans were set to 0 degrees. The γ-analysis passing rates for the whole plan and each beam were gotten using the different standards in the 3 groups, The results showed the mean passing rate of γ-analysis was highest in the P(o) group, and the mean passing rate of the whole plan was better than that of each beam. In addition, it became worse with the increase of dose leakage between the leaves in P(y) group. Therefore, the determination of isocenter has a visible effect for IMRT dose verification of the 2D chamber array, The isocenter of the planning design should be close to the geometric center of target.

Variation of patient imaging doses with scanning parameters for linac-integrated kilovoltage cone beam CT.

To evaluate the Elekta kilovoltage CBCT doses and the associated technical protocols with patient dosimetry estimation. Image guidance technique with cone-beam CT (CBCT) in radiation oncology on a daily basis can deliver a significant dose to the patient. To evaluate the patient dose from LINAC-integrated kV cone beam CT imaging in image-guided radiotherapy. CT dose index (CTDI) were measured with PTW TM30009 CT ion chamber in air, in head phantom and body phantom, respectively; with different combinations of tube voltage, current, exposure time per frame, collimator and gantry rotation range. Dose length products (DLP) were subsequently calculated to account for volume integration effects. The CTDI and DLP were also compared to AcQSim™ simulator CT for routine clinical protocols. Both CTDIair and CTDIw depended quadratically on the voltage, while linearly on milliampere x seconds (mAs) settings. It was shown that CTDIw and DLP had very close relationship with the collimator settings and the gantry rotation ranges. Normalized CTDIw for Elekta XVI™ CBCT was lower than that of ACQSim simulator CT owing to its pulsed radiation output characteristics. CTDIw can be used to assess the patient dose in CBCT due to its simplicity for measurement and reproducibility. Regular measurement should be performed in QA & QC program. Optimal image parameters should be chosen to reduce patient dose during CBCT.

[Patient-specific dose verification method using ArcCHECK for total marrow irradiation with intensity modulated arc therapy].

To investigate the patient-specific dose verification method using ArcCHECK for total marrow irradiation (TMI) with Volumetric Modulated Arc Therapy (VMAT) and Helical Tomotherapy (HT). The kVCT images collected from 8 patients were respectively designed for RapidArc and Tomotherapy plans in total marrow irradiation. ArcCHECK was used for dose verification for the head-neck, chest-abdomen and pelvic. The merging function of ArcCHECK was used in VMAT and the method of double plans (reference and delivery plans) were used in HT. The γ-analysis passing rates for the head-neck, chest-abdomen, pelvic were 98.9% ± 1.9%, 98.4% ± 1.8%, 97.4% ± 2.1% for VMAT plans and 94.3% ± 1.5%, 96.5 ± 1.2%, 94.1% ± 1.9% for HT plans. The results show that using the merging function of ArcCHECK can achieve the dose verification well for VMAT plans with TMI. The method of double plans was done for the dose verification of HT plans with TMI as well as the plans with the targets keeping away from the set-up center.

A Dosimetric Analysis of Preoperative Intensity-modulated and Image-guided Radiation Therapy with and without Simultaneous Integrated Boost for Locally Advanced Rectal Cancer.

Preoperative concurrent chemoradiation, total mesorectal excision and adjuvant chemotherapy have become the standard of care for patients with locally advanced rectal cancer (LARC). Several studies have reported increased pathologic complete response rates and improved locoregional control with escalating doses of preoperative radiotherapy. In this study, we assess the dosimetric feasibility and impact of intensity-modulated and image-guided radiation therapy (IMRT-IGRT) with a simultaneous integrated boost (SIB) in preoperative chemoradiation for LARC. Ten rectal cancer patients treated with preoperative chemoradiation were enrolled in this study, and IMRT56.25Gy and IMRT50Gy plans were made for each patient with a CTV-PTV50Gy margin of 5 mm and a GTV-PTV56.25Gy margin of 10 mm adapted to daily KV cone-beam computed tomography (CBCT) imaging. In the boost group (IMRT56.25Gy), the prescribed doses were 56.25 Gy to the gross tumor (PTV56.25Gy) and 50 Gy to areas at high risk of harboring microscopic disease (PTV50Gy). Doses were delivered over 25 daily fractions using a SIB technique. In the no-boost group (IMRT50Gy), the prescribed dose was 50 Gy to PTV50Gy without a boost. The goals were to give at least 95% of the prescribed doses to at least 95% of the PTVs while keeping irradiated volumes of the organs at risk dose as low as possible. Differences in dose distributions between the two sets of plans were analyzed using a paired sample t-test. All IMRT56.25Gy plans met the needs of the prescribed doses and organ at risk dose constraints. Compared to IMRT50Gy, the addition of a SIB in IMRT56.25Gy resulted in significant increases in mean dose and V40Gy to the bladder and significant increases of V30Gy and V40Gy to femoral heads (p < 0.05 for all points). There were no significant differences in dose to small bowel or pelvic bone marrow between the two sets of plans. Preoperative IMRT-IGRT with SIB for LARC is feasible dosimetrically with respect to organ at risk dose constraints. A phase II trial to evaluate the clinical safety and efficacy of this approach is being undertaken.

[Radiation output evaluation of kilovoltage cone beam CT unit].

OBJECTIVE: To evaluate the radiation output and stability of linac-integrated kV cone beam CT unit. METHODS: Air kermas in radiographic mode were measured with 0.6 cc ion chamber and Unidos electrometer for Synergy-integrated XVI kV cone beam CT unit. Air kermas vs image frames were measured in fluoroscopic mode. Output stability and depth doses were measured. RESULTS: The air kerma increased quadratically with the increased tube voltage, while increasing linearly with the tube current, exposure time, and number of frames. The radiation output stability and its change with the gantry angle were within +/-1%. The percentage depth dose increased with higher tube voltage. CONCLUSION: The radiation output of XVI is stable. The radiation outputs change considerably with the preset parameters. Parameters should be optimally chosen to reduce the patient dose.

[Analysis of megavoltage computed tomography imaging on a helical tomotherapy unit].

OBJECTIVE: To evaluate the image quality of megavoltage computed Tomography imaging. METHODS: The HU uniformity and linearity, image noise, spatial resolution, low contrast resolution and spatial linearity in MVCT mode were evaluated with Catphan 600 phantom, and the factor of pitch was also evaluated. Influencing factors of image quality were also discussed. RESULTS: The MVCT values depended linearly on the physical density of the sample. The MVCT values uniformity was good. The spatial resolution was 4 lp/cm. The use of an MV Beam for imaging results in the loss of low contrast resolution, but it is sufficient for pretreatment image guidance. The geometric accuracy was good. CONCLUSIONS: The image quality of MVCT is less than that of KVCT, but is good enough for IGRT.