To investigate the feasibility of synthesizing computed tomography (CT) images from magnetic resonance (MR) images using generative adversarial networks (GANs) for nasopharyngeal carcinoma (NPC) intensity-modulated radiotherapy (IMRT) planning. Conventional T1-weighted MR images and CT images were acquired from 173 NPC patients. The MR and CT images of 28 patients were randomly chosen as the independent tested set. The remaining images were used to build a conditional GAN (cGAN) and a cycle-consistency GAN (cycleGAN). A U-net was used as the generator in cGAN, whereas a residual-Unet was used as the generator in cycleGAN. The cGAN was trained using the deformable registered MR-CT image pairs, whereas the cycleGAN was trained using the unregistered MR and CT images. The generated synthetic CT (SCT) images from cGAN and cycleGAN were compared with the true CT images with respect to their Hounsfield Unit (HU) discrepancy and dosimetric accuracy for NPC IMRT plans. The mean absolute errors within the body were 69.67 ± 9.27 HU and 100.62 ± 7.39 HU for the cGAN and cycleGAN, respectively. The 2%/2-mm γ passing rates were (98.68 ± 0.94)% and (98.52 ± 1.13)% for the cGAN and cycleGAN, respectively. Meanwhile, the absolute dose discrepancies within the regions of interest were (0.49 ± 0.24)% and (0.62 ± 0.36)%, respectively. Both cGAN and cycleGAN could swiftly generate accurate SCT volume images from MR images, with high dosimetric accuracy for NPC IMRT planning. cGAN was preferable if high-quality MR-CT image pairs were available.

The purpose of this study was to propose and evaluate a method of creating a synthetic CT (S-CT) from MRI simulation for dose calculation and daily CBCT localization. A pair of MR and CT images was obtained in the same day from each of 10 prostate patients. The pair of MR and CT images was preregistered using the deformable image registration (DIR). Using the corresponding displacement vector field (atlas-DVF), the CT image was deformed to the MR image to create an atlas MR-CT pair. Regions of interest (ROI) on the atlas MR-CT pair were delineated and used to create atlas-ROI masks. ‘Leave-one-out’ test (one pair of MR and CT was used as subject-MR and subject-CT for evaluation, and the remaining 9 pairs were in the atlas library) was performed. For a subject-MR, autosegmentation and DVFs were generated using DIR between the subject-MR and the 9 atlas-MRs. An S-CT was then generated using the corresponding 9 paired atlas-CTs, the 9 atlas-DVFs and the corresponding atlas-ROI masks. The total 10 S-CTs were evaluated using the Hounsfield unit (HU), the calculated dose distribution, and the auto bony registration to daily CBCT images with respect to the 10 subject-CTs. HU differences (\text{mean}?\text{STD}) were (2.4?25.23) , (1.18?39.49) , (32.46?81.9) , (0.23?40.13) , and (3.74?144.76) for prostate, bladder, rectal wall, soft tissue outside all ROIs, and bone, respectively. The discrepancy of dose-volume parameters calculated using the S-CT for treatment planning was small ( ?0.22\% with 95% confidence). Gamma pass rate (2% & 2 mm) was higher than 99.86% inside PTV and 98.45% inside normal structures. Using the 10 S-CTs as the reference CT for daily CBCT localization achieved the similar results compared to using the subject-CT. The translational vector differences were within 1.08 mm (0.37?0.23?\text{mm}) , and the rotational differences were within 1.1° in all three directions. S-CT created from a simulation MR image using the proposed approach with the preconstructed atlas library can replace the planning CT for dose calculation and daily CBCT image guidance. PACS number(s): 87.57.nf, 87.57.nm

To construct a tumor voxel dose response matrix (DRM) and dose prescription function (DPF) for adaptive dose painting by number (DPbN) based on treatment feedback of fluoro-2-deoxyglucose (FGD) positron emission tomography (PET)/computed tomography (CT) imaging. FDG-PET/CT images obtained before and after chemoradiation therapy and at weekly chemoradiation therapy sessions for each of 18 patients with head and neck cancer, as well as the treatment outcomes, were used in the modeling. All weekly and posttreatment PET/CT images were registered voxel-to-voxel to the corresponding pretreatment baseline PET/CT image. Tumor voxel DRM was created using serial FDG-PET imaging of each patient with respect to the baseline standardized uptake value (SUV0). A tumor voxel control probability (TVCP) lookup table was created using the maximum likelihood estimation on the tumor voxel (SUV0, DRM) domain of all tumors. Tumor voxel DPF was created from the TVCP lookup table and used as the objective function for DPbN-based inverse planning optimization. Large intertumoral and intratumoral variations on both tumor voxels (SUV0, DRM) were identified. Tumor voxel dose resistance did not show correlation with its baseline SUV0 value and was the major cause of the tumor local failures. Tumor voxel DPF as the function of tumor voxel (SUV0, DRM) values also showed a very large intertumoral and intratumoral heterogeneity. Most human papillomavirus-negative tumors require a treatment dose >100 Gy to certain local tumor regions. These treatment doses, which are most unlikely to be implementable in conventional radiation therapy, can be achieved using adaptive DPbN treatment. Clinical feasibility was evaluated by comparing the adaptive DPbN treatment plan with the conventional intensity modulated radiation therapy plan. Tumor voxel (SUV0, DRM) provides an intratumoral prognostic map to target tumor locoregional-resistant regions. The corresponding TVCP or DPF provides a quantitative objective to optimize the intratumoral dose distribution for the individuals. The adaptive DPbN with FDG-PET/CT imaging feedback is feasible to implement in clinics.

To quantify the dosimetric parameters of different bone marrow sparing strategies and to determine the optimal strategy for cervical cancer patients undergoing postoperative intensity-modulated radiotherapy (IMRT). Fifteen patients with cervical cancer were selected for analysis. The planning target volume (PTV) and the organs at risks (OAR) including small bowel, bladder, rectum, femoral heads, os coxae (OC), lumbosacral spine (LS) and bone marrow (BM) were contoured. For each patient, four IMRT plans with different strategies were generated, including one plan without BM as the dose-volume constraint, namely IMRT (N) plan, and three bone marrow sparing (BMS-IMRT) plans. The three BMS-IMRT plans used the BM, OC, OC and LS respectively, as the BM OAR, namely as IMRT (BM), IMRT (OC) and IMRT (OC + LS) plans. Dose volumes for the target and the OARs were compared using analysis of variance (ANOVA). Compared with IMRT (N) plans, the dose to the small bowel, bladder, rectum and femoral heads showed no increase in the three BMS-IMRT plans. However, the irradiated dose to BM, OC and LS significantly decreased. In particular, the mean dose of BM, OC and LS decreased by about 5Gy (p < 0.05) in IMRT (BM) plans while the average volume receiving ≥20, ≥30, ≥40Gy decreased by 7.1–24.2%. The LS volume receiving 40Gy showed the highest decrease (about 31.2%, p < 0.05) in IMRT (OC + LS) plans. On the other hand, in comparison with IMRT (BM), IMRT (OC) reduced the dose volume of to the OC, but increased the dose to LS while IMRT (OC + LS) plans reduced both the OC and the LS volume at all dose levels. Specifically, the V20 of OC and LS in the IMRT (OC + LS) plan decreased by 11.5 and 11.2%, respectively. By introducing the os coxae and lumbosacral spine as the dose–volume constraints, the IMRT plans exhibited the best sparing of the bone marrow without compromising the dose to surrounding normal structures. Therefore, we recommend adding the os coxae and lumbosacral spine as the BM OAR in such plans.

To develop an advanced deep convolutional neural network (DCNN) architecture to generate synthetic CT (SCT) images from MR images for intensity-modulated proton therapy (IMPT) treatment planning of...