Deep Learning (DL) technique has shown great potential but still has limited success in online contouring for MR-guided adaptive radiotherapy (MRgART). This study proposed a patient-specific DL auto-segmentation (DLAS) strategy using the patient's previous images and contours to update the model and improve segmentation accuracy and efficiency for MRgART. A prototype model was trained for each patient using the first set of MRI and corresponding contours as inputs. The patient-specific model was updated after each fraction with all the available fractional MRIs/contours, and then used to predict the segmentation for the next fraction. During model training, a variant was fitted under consistency constraints, limiting the differences in the volume, length and centroid between the predictions for the latest MRI within a reasonable range. The model performance was evaluated for both organ-at-risks and tumors auto-segmentation for a total of 6 abdominal/pelvic cases (each with at least 8 sets of MRIs/contours) underwent MRgART through Dice Similarity Coefficient (DSC) and 95% Hausdorff Distance (HD95), and was compared with deformable image registration (DIR) and frozen DL model (no updating after pre-training). The contouring time was also recorded and analyzed. The proposed model achieved superior performance with higher mean DSC (0.90, 95 % CI: 0.88-0.95), as compared to DIR (0.63, 95 %CI: 0.59-0.68) and frozen DL models (0.74, 95 % CI: 0.71-0.79). As for tumors, the proposed method yielded a median DSC of 0.95, 95 % CI: 0.94-0.97, and a median HD95 of 1.63 mm, 95 % CI: 1.22 mm-2.06 mm. The contouring time was reduced significantly (p < 0.05) using the proposed method (73.4 ± 6.5 secs) compared to the manual process (12 ∼ 22 mins). The online ART time was reduced to 1650 ± 274 seconds with the proposed method, as compared to 3251.8 ± 447 seconds using the original workflow. The proposed patient-specific DLAS method can significantly improve the segmentation accuracy and efficiency for longitudinal MRIs, thereby facilitating the routine practice of MRgART.

Purpose: This study aimed to assess the effects of bladder filling during cervical cancer radiotherapy on target volume and organs at risk (OARs) dose based on daily computed tomography (daily‐CT) images and provide bladder‐volume‐based dose prediction models.Methods: Nineteen patients (475 daily‐CTs) comprised the study group, and five patients comprised the validation set (25 daily‐CTs). Target volumes and OARs were delineated on daily‐CT images and the treatment plan was recalculated accordingly. The deviation from the planning bladder volume (DVB), the correlation between DVB and clinical (CTV)/planning (PTV) target volume in terms of prescribed dose coverage, and the relationship of small bowel volume and bladder dose with the ratio of bladder volume (RVB) were analyzed.Results: In all cases, the prescribed dose coverage in the CTV was >95% when DVB was 95% when RVB was <160%. The ratio of bladder V_{45 Gy} to the planning bladder V_{45 Gy} (RBV₄₅) exhibited a negative linear relationship with RVB (RBV₄₅ = −0.18*RVB + 120.8; R² = 0.80). Moreover, the ratio of small bowel volume to planning small bowel volume (RVS) exhibited a negative linear relationship with RVB (RVS = −1.06*RVB +217.59; R² = 0.41). The validation set results showed that the linear model predicted well the effects of bladder volume changes on target volume coverage and bladder dose.Conclusions: This study assessed dosimetry and volume effects of bladder filling on target and OARs based on daily‐CT images. We established a quantitative relationship between these parameters, providing dose prediction models for cervical cancer radiotherapy.

Conventional phantom-based planar dosimetry (2D-PBD) quality assurance (QA) using gamma pass rate (GP (%)) is inadequate to reflect clinically relevant dose error in intensity-modulated radiation therapy (IMRT), owing to a lack of information regarding patient anatomy and volumetric dose distribution. This study aimed to evaluate the dose distribution accuracy of IMRT delivery for nasopharyngeal carcinoma (NPC), which passed the 2D-PBD verification, using a measurement-guided 3D dose reconstruction (3D-MGR) method. Radiation treatment plans of 30 NPC cases and their pre-treatment 2D-PBD data were analyzed. 3D dose distribution was reconstructed on patient computed tomography (CT) images using the 3DVH software and compared to the treatment plans. Global and organ-specific dose GP (%), and dose-volume histogram (DVH) deviation of each structure was evaluated. Interdependency between GP (%) and the deviation of the volumetric dose was studied through correlation analysis. The 3D-MGR achieved global GP (%) similar to conventional 2D-PBD in the same criteria. However, structure-specific GP (%) significantly decreased under stricter criteria, including the planning target volume (PTV). The average deviation of all inspected dose volumes (DV) and volumetric dose (VD) parameters ranged from − 2.93% to 1.17%, with the largest negative deviation in V100% of the PTVnx of − 15.66% and positive deviation in D1cc of the spinal cord of 6.66%. There was no significant correlation between global GP (%) of 2D-PBD or 3D-MGR and the deviation of the most volumetric dosimetry parameters (DV or VD), when the Pearson’s coefficient value of 0.8 was used for correlation evaluation. Even upon passing the pre-treatment phantom based dosimetric QA, there could still be risk of dose error like under-dose in PTVnx and overdose in critical structures. Measurement-guided 3D volumetric dosimetry QA is recommended as the more clinically efficient verification for the complicated NPC IMRT.

Purpose: We aimed to perform the commissioning and clinical evaluation of myQA SRS detector array for patient‐specific quality assurance (PSQA) of stereotactic radiosurgery (SRS)/ stereotactic body radiotherapy (SBRT) plans.Methods: To perform the commissioning of myQA SRS, its dose linearity, dose‐rate dependence, angular dependence, and field‐size dependence were investigated. Ten SBRT plans were selected for clinical evaluation: 1) Common clinical deviations based on the original SBRT plan (Plan0), including multileaf collimator (MLC) positioning deviation and treatment positioning deviation were introduced. 2) Compared the performance of the myQA SRS and a high‐resolution EPID dosimetry system in PSQA measurement for the SBRT plans. Evaluation parameters include gamma passing rate (GPR) and distance‐to‐agreement (DTA) pass rate (DPR).Results: The dose linearity, angle dependence, and field‐size dependence of myQA SRS system exhibit excellent performance. The myQA SRS is highly sensitive in the detection of MLC deviations. The GPR of (3%/1 mm) decreases from 90.4% of the original plan to 72.7%/62.9% with an MLC outward/inward deviation of 3 mm. Additionally, when the setup error deviates by 1 mm in the X, Y, and Z directions with the GPR of (3%/1 mm) decreasing by an average of −20.9%, −25.7%, and −24.7%, respectively, and DPR (1 mm) decreasing by an average of −33.7%, −32.9%, and −29.8%. Additionally, the myQA SRS has a slightly higher GPR than EPID for PSQA, However, the difference is not statistically significant with the GPR of (3%/1 mm) of (average 90.4%% vs. 90.1%, p = 0.414).Conclusion: Dosimetry characteristics of the myQA SRS device meets the accuracy and sensitivity requirement of PSQA for SRS/SBRT treatment. The dose rate dependence should be adequately calibrated before its application and a more stringent GPR (3%/1 mm) evaluation criterion is suggested when it is used for SRS/SBRT QA.

Purpose: To assist in the selection of a suitable combination of an irradiation technique and jaw condition in intensity‐modulated radiotherapy (IMRT) and volumetric‐modulated arc radiotherapy (VMAT) for lung cancer treatment plans.Materials and methods: Thirty patients with lung cancer who underwent radiotherapy were enrolled retrospectively. They were categorized as having central lung cancer, peripheral lung cancer with mediastinal lymph node metastasis (peripheral E lung cancer), and peripheral lung cancer without mediastinal lymph node metastasis (peripheral N lung cancer). Four treatment plans were designed for each patient: fixed jaw and adaptive jaw IMRT technique (FJ‐IMRT and JA‐IMRT), and fixed jaw and jaw tracking VMAT technique (FJ‐VMAT and JT‐VMAT). The dose parameters of the four group plans were compared and analyzed.Results: Compared to FJ‐IMRT, JA‐IMRT significantly reduced the mean dose (D_mean) and volume percentage of 5 Gy (V_5Gy) of the total lung in central and peripheral N lung cancer. Similarly, compared to FJ‐VMAT, JT‐VMAT provided better protection to most organs at risk (OARs), particularly for total lung and heart. In comparison with IMRT, VMAT significantly improved the conformity index (CI) of the planning target volume for the three lung cancer classifications, and it reduced the dose of almost all OARs except V_5Gy and D_mean of the total lung. Moreover, the mean monitor units of the VMAT groups were far lower than the IMRT groups.Conclusion: Based on the dosimetric findings and considering clinical data published on lung and heart side effects, we propose recommendations on the preferred treatment technique based on tumor location and pulmonary function. For central lung cancer with normal pulmonary function, we advise JT‐VMAT techniques. Conversely, for central lung cancer with poor pulmonary function, we recommend JA‐IMRT techniques. We advocate JA‐IMRT for peripheral E lung cancer. For peripheral N lung cancer, JT‐VMAT techniques are strongly recommended.

To investigate the effect of different MR sequences on the accuracy of Cycle-consistent Generative Adversarial Network (CycleGAN) based synthetic CT (sCT) generation in nasopharyngeal carcinoma (NPC). In this work, three sequences of MR, included T1-weighted imaging (T1WI), T2-weighted imaging (T2WI), and T1 contrast-enhanced weighted imaging (T1WIC), and planning CT (PCT) images of 151 patients with NPC were collected. The number of training, verification, and test sets were 108, 16, and 27, respectively. Three unsupervised CycleGAN-based models were trained using different sequences (single channels) as inputs, and the synergistic fusion model were used multiple channels. To assess the precision of these models, we evaluated key metrics such as mean error (ME), mean absolute error (MAE), structural similarity index (SSIM), peak signal-to-noise ratio (PSNR), and dose distribution, comparing the PCT with the sCT generated by each model. Overall, The SCT _T2 image generated by T2WI model achieved superior results than those of T1WIC and T1WI for a single-sequence model. Compared to the SCT _T2 , the quality of the SCT _Fusion image generated by the fusion model was further improved, with significant differences in the MAE, SSIM, and PNSR metrics (P<0.05), the average MAE, SSIM, and PNSR in the body were (112.97±9.69 vs. 90.99±9.64, P<0.05) HU, (0.86±0.02 vs. 0.89±0.02, P<0.05), and (21.66±0.64 vs. 23.28±0.87, P<0.05), respectively. Meanwhile, the average gamma passing rates (3%, 3 mm) and the average absolute dose discrepancies were 99.45% ± 1.05% and 0.73%±0.73% for the fusion model. In conclusion, our findings reveal that the CycleGAN model, particularly when employing fusion MR sequences as input, offers the highest accuracy in synthetic CT generation. Notably, T2WI images stand out as a viable option for sCT prediction in clinical settings where acquisition sequences or times are limited.