Channel-wise attention enhanced and structural similarity constrained cycleGAN for effective synthetic CT generation from head and neck MRI images.

BACKGROUND: Magnetic resonance imaging (MRI) plays an increasingly important role in radiotherapy, enhancing the accuracy of target and organs at risk delineation, but the absence of electron density information limits its further clinical application. Therefore, the aim of this study is to develop and evaluate a novel unsupervised network (cycleSimulationGAN) for unpaired MR-to-CT synthesis. METHODS: The proposed cycleSimulationGAN in this work integrates contour consistency loss function and channel-wise attention mechanism to synthesize high-quality CT-like images. Specially, the proposed cycleSimulationGAN constrains the structural similarity between the synthetic and input images for better structural retention characteristics. Additionally, we propose to equip a novel channel-wise attention mechanism based on the traditional generator of GAN to enhance the feature representation capability of deep network and extract more effective features. The mean absolute error (MAE) of Hounsfield Units (HU), peak signal-to-noise ratio (PSNR), root-mean-square error (RMSE) and structural similarity index (SSIM) were calculated between synthetic CT (sCT) and ground truth (GT) CT images to quantify the overall sCT performance. RESULTS: One hundred and sixty nasopharyngeal carcinoma (NPC) patients who underwent volumetric-modulated arc radiotherapy (VMAT) were enrolled in this study. The generated sCT of our method were more consistent with the GT compared with other methods in terms of visual inspection. The average MAE, RMSE, PSNR, and SSIM calculated over twenty patients were 61.88 ± 1.42, 116.85 ± 3.42, 36.23 ± 0.52 and 0.985 ± 0.002 for the proposed method. The four image quality assessment metrics were significantly improved by our approach compared to conventional cycleGAN, the proposed cycleSimulationGAN produces significantly better synthetic results except for SSIM in bone. CONCLUSIONS: We developed a novel cycleSimulationGAN model that can effectively create sCT images, making them comparable to GT images, which could potentially benefit the MRI-based treatment planning.

TransQA: deep hybrid transformer network for measurement-guided volumetric dose prediction of pre-treatment patient-specific quality assurance.

Objective. Performing pre-treatment patient-specific quality assurance (prePSQA) is considered an essential, time-consuming, and resource-intensive task for volumetric modulated arc radiotherapy (VMAT) which confirms the dose accuracy and ensure patient safety. Most current machine learning and deep learning approaches stack excessive convolutional/pooling operations (CPs) to predict prePSQA with two-dimensional or one-dimensional information input. However, these models generally present limitations in explicitly modeling long-range dependency for volumetric dose prediction due to the loss of spatial dose features and the inherent locality of CPs. The purpose of this work is to construct a deep hybrid network by combining the self-attention mechanism-based Transformer with modified U-Net for predicting measurement-guided volumetric dose (MDose) of prePSQA.Approach. The enrolled 307 cancer patients underwent VMAT were randomly divided into 246 and 61 cases for training and testing the model. The input data included computed tomography images, radiotherapy dose images exported from the treatment planning system, as well as the MDose distribution from the verification system. The output was the predicted high-quality voxel-wise prePSQA dose distribution.Main results: qualitative and quantitative experimental results show that the proposed prediction method could achieve comparable or better performance on MDose prediction over other approaches in terms of spatial dose distribution, dose-volume histogram metrics, gamma passing rates, mean absolute error, root mean square error, and structural similarity.Significance. The preliminary results on multiple cancer sites show that our approach can be taken as a clinical guidance tool and help medical physicists to reduce the measurement work of prePSQA.

Comparative dosimetric and radiobiological assessment of left-sided whole breast and regional nodes with advanced radiotherapy techniques.

The aim of this study was to analyze the dosimetric and radiobiologic differences of the left-sided whole breast and regional nodes in intensity-modulated radiotherapy (IMRT), volume-modulated arc therapy (VMAT), and helical tomotherapy (HT).  The IMRT, VMAT, and HT plans in this study were generated for thirty-five left-sided breast cancer patients after breast-conserving surgery (BCS). The planning target volume (PTV) included the whole breast and supraclavicular nodes. PTV coverage, homogeneity index (HI), conformity index (CI), dose to organs at risk (OARs), secondary cancer complication probability (SCCP), and excess absolute risk (EAR) were used to evaluate the plans. Compared to IMRT, the VMAT and HT plans resulted in higher PTV coverage and homogeneity. The VMAT and HT plans also delivered a lower mean dose to the ipsilateral lung (9.19 ± 1.36 Gy, 9.48 ± 1.17 Gy vs. 11.31 ± 1.42 Gy) and heart (3.99 ± 0.86 Gy, 4.48 ± 0.62 Gy vs. 5.53 ± 1.02 Gy) and reduced the V5Gy, V10Gy, V20Gy, V30Gy, and V40Gy of the ipsilateral lung and heart. The SCCP and EAR for the ipsilateral lung were reduced by 3.67%, 3.09% in VMAT, and 22.18%, 19.21% in HT, respectively. While were increased for the contralateral lung and breast. This study showed that VMAT plans provide a more homogeneous dose distribution to the PTV, minimizing exposure to ipsilateral structures and significantly reducing SCCP and EAR, and slightly increasing dose to contralateral structures. Overall, the VMAT plan can be considered a beneficial technique for BCS patients whose PTV includes the whole breast and regional nodes.

Multi-isocenter VMAT craniospinal irradiation using feasibility dose-volume histogram-guided auto-planning technique.

This study aims to propose a novel treatment planning methodology for multi-isocenter volumetric modulated arc therapy (VMAT) craniospinal irradiation (CSI) using the special feasibility dose-volume histogram (FDVH)-guided auto-planning (AP) technique. Three different multi-isocenter VMAT -CSI plans were created, including manually based plans (MUPs), conventional AP plans (CAPs) and FDVH-guided AP plans (FAPs). The CAPs and FAPs were specially designed by combining multi-isocenter VMAT and AP techniques in the Pinnacle treatment planning system. Specially, the personalized optimization parameters for FAPs were generated using the FDVH function implemented in PlanIQ software, which provides the ideal organs at risk (OARs) sparing for the specific anatomical geometry based on the valuable assumption of the dose fall-off. Compared to MUPs, CAPs and FAPs significantly reduced the dose for most of the OARs. FAPs achieved the best homogeneity index (0.092 ± 0.013) and conformity index (0.980 ± 0.011), while CAPs were slightly inferior to the FAPs but superior to the MUPs. As opposed to MUPs, FAPs delivered a lower dose to OARs, whereas the difference between FAPs and CAPs was not statistically significant except for the optic chiasm and inner ear_L. The two AP approaches had similar MUs, which were significantly lower than the MUPs. The planning time of FAPs (145.00 ± 10.25 min) was slightly lower than that of CAPs (149.83 ± 14.37 min) and was substantially lower than that of MUPs (157.92 ± 16.11 min) with P < 0.0167. Overall, introducing the multi-isocenter AP technique into VMAT-CSI yielded positive outcomes and may play an important role in clinical CSI planning in the future.

Generating synthesized computed tomography from CBCT using a conditional generative adversarial network for head and neck cancer patients.

Purpose: To overcome the imaging artifacts and Hounsfield unit inaccuracy limitations of cone-beam computed tomography, a conditional generative adversarial network is proposed to synthesize high-quality computed tomography-like images from cone-beam computed tomography images. Methods: A total of 120 paired cone-beam computed tomography and computed tomography scans of patients with head and neck cancer who were treated during January 2019 and December 2020 retrospectively collected; the scans of 90 patients were assembled into training and validation datasets, and the scans of 30 patients were used in testing datasets. The proposed method integrates a U-Net backbone architecture with residual blocks into a conditional generative adversarial network framework to learn a mapping from cone-beam computed tomography images to pair planning computed tomography images. The mean absolute error, root-mean-square error, structural similarity index, and peak signal-to-noise ratio were used to assess the performance of this method compared with U-Net and CycleGAN. Results: The synthesized computed tomography images produced by the conditional generative adversarial network were visually similar to planning computed tomography images. The mean absolute error, root-mean-square error, structural similarity index, and peak signal-to-noise ratio calculated from test images generated by conditional generative adversarial network were all significantly different than CycleGAN and U-Net. The mean absolute error, root-mean-square error, structural similarity index, and peak signal-to-noise ratio values between the synthesized computed tomography and the reference computed tomography were 16.75 ± 11.07 Hounsfield unit, 58.15 ± 28.64 Hounsfield unit, 0.92 ± 0.04, and 30.58 ± 3.86 dB in conditional generative adversarial network, 20.66 ± 12.15 Hounsfield unit, 66.53 ± 29.73 Hounsfield unit, 0.90 ± 0.05, and 29.29 ± 3.49 dB in CycleGAN, and 16.82 ± 10.99 Hounsfield unit, 58.68 ± 28.34 Hounsfield unit, 0.92 ± 0.04, and 30.48 ± 3.83 dB in U-Net, respectively. Conclusions: The synthesized computed tomography generated from the cone-beam computed tomography-based conditional generative adversarial network method has accurate computed tomography numbers while keeping the same anatomical structure as cone-beam computed tomography. It can be used effectively for quantitative applications in radiotherapy.

Comparison of an in-house hybrid DIR method to NiftyReg on CBCT and CT images for head and neck cancer.

An in-house hybrid deformable image registration (DIR) method, which combines free-form deformation (FFD) and the viscous fluid registration method, is proposed. Its results on the planning computed tomography (CT) and the day 1 treatment cone-beam CT (CBCT) image from 68 head and neck cancer patients are compared with the results of NiftyReg, which uses B-spline FFD alone. Several similarity metrics, the target registration error (TRE) of annotated points, as well as the Dice similarity coefficient (DSC) and Hausdorff distance (HD) of the propagated organs at risk are employed to analyze their registration accuracy. According to quantitative analysis on mutual information, normalized cross-correlation, and the absolute pixel value differences, the results of the proposed DIR are more similar to the CBCT images than the NiftyReg results. Smaller TRE of the annotated points is observed in the proposed method, and the overall mean TRE for the proposed method and NiftyReg was 2.34 and 2.98 mm, respectively (p < 0.001). The mean DSC in the larynx, spinal cord, oral cavity, mandible, and parotid given by the proposed method ranged from 0.78 to 0.91, significantly higher than the NiftyReg results (ranging from 0.77 to 0.90), and the HD was significantly lower compared to NiftyReg. Furthermore, the proposed method did not suffer from unrealistic deformations as the NiftyReg did in the visual evaluation. Meanwhile, the execution time of the proposed method was much higher than NiftyReg (96.98 ± 11.88 s vs. 4.60 ± 0.49 s). In conclusion, the in-house hybrid method gave better accuracy and more stable performance than NiftyReg.

A clinical trial to compare a 3D-printed bolus with a conventional bolus with the aim of reducing cardiopulmonary exposure in postmastectomy patients with volumetric modulated arc therapy.

BACKGROUND: We compared the dosimetry, application, and acute toxicity of a 3D-printed and a conventional bolus for postmastectomy radiotherapy (PMRT) with volumetric modulated arc therapy (VMAT). Materials and Methods Eligible patients (n = 75) with PMRT breast cancer were randomly selected to receive VMAT with a conventional bolus or a 3D-printed bolus. The primary endpoint was a 10% decrease in the mean heart dose to left-sided breast cancer patients. The secondary endpoint was a 5% decrease in the mean ipsilateral lung dose to all patients. A comparative analysis was carried out of the dosimetry, normal tissue complication probability (NTCP), acute skin toxicity, and radiation pneumonitis. RESULTS: Compared to a conventional bolus, the mean heart dose in left-sided breast cancer was reduced by an average of 0.8 Gy (5.5 ± 1.3 Gy vs. 4.7 ± 0.8 Gy, p = 0.035) and the mean dose to the ipsilateral lung was also reduced by an average of 0.8 Gy (12.4 ± 1.0 Gy vs. 11.6 ± 0.8 Gy, p < 0.001). The values for V50Gy of the PTV of the chest wall for the 3D-printed and conventional boluses were 95.4 ± 0.6% and 94.8 ± 0.8% (p = 0.026) and the values for the CI of the entire PTV were 0.83 ± 0.02 and 0.80 ± 0.03 (p < 0.001), respectively. The NTCP for the 3D-printed bolus was also reduced to an average of 0.14% (0.32 ± 0.19% vs. 0.18 ± 0.11%, p = 0.017) for the heart and 0.45% (3.70 ± 0.67% vs. 3.25 ± 0.18%, p < 0.001) for the ipsilateral lung. Grade 2 and Grade 1 radiation pneumonitis were 0.0% versus 7.5% and 14.3% versus 20.0%, respectively (p = 0.184). CONCLUSIONS: The 3D-printed bolus may reduce cardiopulmonary exposure in postmastectomy patients with volumetric modulated arc therapy.

A dosimetric and radiobiological evaluation of VMAT following mastectomy for patients with left-sided breast cancer.

BACKGROUND: To compare the dosimetric, normal tissue complication probability (NTCP), secondary cancer complication probabilities (SCCP), and excess absolute risk (EAR) differences of volumetric modulated arc therapy (VMAT) and intensity-modulated radiation therapy (IMRT) for left-sided breast cancer after mastectomy. METHODS AND MATERIALS: Thirty patients with left-sided breast cancer treated with post-mastectomy radiation therapy (PMRT) were randomly enrolled in this study. Both IMRT and VMAT treatment plans were created for each patient. Planning target volume (PTV) doses for the chest wall and internal mammary nodes, PTV1, and PTV of the supraclavicular nodes, PTV2, of 50 Gy were prescribed in 25 fractions. The plans were evaluated based on PTV1 and PTV2 coverage, homogeneity index (HI), conformity index, conformity number (CN), dose to organs at risk, NTCP, SCCP, EAR, number of monitors units, and beam delivery time. RESULTS: VMAT resulted in more homogeneous chest wall coverage than did IMRT. The percent volume of PTV1 that received the prescribed dose of VMRT and IMRT was 95.9 ± 1.2% and 94.5 ± 1.6%, respectively (p < 0.001). The HI was 0.11 ± 0.01 for VMAT and 0.12 ± 0.02 for IMRT, respectively (p = 0.001). The VMAT plan had better conformity (CN: 0.84 ± 0.02 vs. 0.78 ± 0.04, p < 0.001) in PTV compared with IMRT. As opposed to IMRT plans, VMAT delivered a lower mean dose to the ipsilateral lung (11.5 Gy vs 12.6 Gy) and heart (5.2 Gy vs 6.0 Gy) and significantly reduced the V5, V10, V20, V30, and V40 of the ipsilateral lung and heart; only the differences in V5 of the ipsilateral lung did not reach statistical significance (p = 0.409). Although the volume of the ipsilateral lung and heart encompassed by the 2.5 Gy isodose line (V2.5) was increased by 6.7% and 7.7% (p < 0.001, p = 0.002), the NTCP was decreased by 0.8% and 0.6%, and SCCP and EAR were decreased by 1.9% and 0.1% for the ipsilateral lung. No significant differences were observed in the contralateral lung/breast V2.5, V5, V10, V20, mean dose, SCCP, and EAR. Finally, VMAT reduced the number of monitor units by 31.5% and the treatment time by 71.4%, as compared with IMRT. CONCLUSIONS: Compared with IMRT, VMAT is the optimal technique for PMRT patients with left-sided breast cancer due to better target coverage, a lower dose delivered, NTCP, SCCP, and EAR to the ipsilateral lung and heart, similar doses delivered to the contralateral lung and breast, fewer monitor units and a shorter delivery time.