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: To construct machine learning models for predicting progression free survival (PFS) and overall survival (OS) with esophageal squamous cell carcinoma (ESCC) patients. Methods: 204 ESCC patients were randomly divided into training cohort (n = 143) and test cohort (n = 61) according to the ratio of 7:3. Two radiomics models were constructed by radiomics features, which were selected by LASSO Cox model to predict PFS and OS, respectively. Clinical features were selected by univariate and multivariate Cox proportional hazards model (p < 0.05). Combined radiomics and clinical model was developed by selected clinical and radiomics features. The receiver operating characteristic curve, Kaplan Meier curve and nomogram were used to display the capability of constructed models. Results: There were 944 radiomics features extracted based on volume of interest in CT images. There were six radiomics features and seven clinical features for PFS prediction and three radiomics features and three clinical features for OS prediction; The radiomics models showed general performance in training cohort and test cohort for prediction for prediction PFS (AUC, 0.664, 0.676. C-index, 0.65, 0.64) and OS (AUC, 0.634, 0.646.C-index, 0.64, 0.65). The combined models displayed high performance in training cohort and test cohort for prediction PFS (AUC, 0.856, 0.833. C-index, 0.81, 0.79) and OS (AUC, 0.742, 0.768. C-index, 0.72, 0.71). Conclusion: We developed combined radiomics and clinical machine learning models with better performance than radiomics or clinical alone, which were used to accurate predict 3 years PFS and OS of non-surgical ESCC patients. The prediction results could provide a reference for clinical decision.

To validate a gradient-based segmentation method for gross tumor volume(GTV) delineation on 8F-fluorothymidine (FLT)positron emission tomography (PET)/ computer tomography (CT) in esophageal squamous cell cancer through pathologic specimen, in comparison with standardized uptake values (SUV) threshold-based methods and CT. The corresponding impact of this GTV delineation method on treatment planning was evaluated. Ten patients with esophageal squamous cell cancer were enrolled. Before radical surgery, all patients underwent FLT-PET/CT. GTVs were delineated by using four methods. GTVGRAD, GTV1.4 and GTV30%max were segmented on FLT PET using a gradient-based method, a fixed threshold of 1.4 SUV and 30% of SUVmax, respectively. GTVCT was based on CT data alone. The maximum longitudinal tumor length of each segmented GTV was compared with the measured tumor length of the pathologic gross tumor length (LPath). GTVGRAD, GTV1.4 and GTV30%max were compared with GTVCT by overlap index. Two radiotherapy plannings (planGRAD) and (planCT) were designed for each patient based on GTVGRAD and GTVCT. The dose-volume parameters for target volume and normal tissues, CI and HI of planGRAD and planCT were compared. The mean ± standard deviation of LPath was 6.47 ± 2.70 cm. The mean ± standard deviation of LGRAD,L1.4, L30%max and LCT were 6.22 ± 2.61, 6.23 ± 2.80, 5.95 ± 2.50,7.17 ± 2.28 cm, respectively. The Pearson correlation coefficients between LPath and each segmentation method were 0.989, 0.920, 0.920 and 0.862, respectively. The overlap indices of GTVGRAD, GTV1.4, GTV30%max when compared with GTVCT were 0.75 ± 0.12, 0.71 ± 0.12, 0.57 ± 0.10, respectively. The V5, V10, V20, V30 and mean dose of total-lung,V30 and mean dose of heart of planGRAD were significantly lower than planCT. The gradient-based method provided the closest estimation of target length. The radiotherapy plannings based on the gradient-based segmentation method reduced the irradiated volume of lung, heart in comparison to CT.

Objective: Halcyon accelerator applies the flattening filter (FF)-free mode instead of the lead gate and FF treatment mode for traditional C-type accelerators. We aimed at comparing and analyzing the quality and delivery of nasopharyngeal carcinoma (NPC) plans between Halcyon and VitalBeam (VB) accelerators in fixed-field intensity-modulated radiation therapy (IMRT). Methods: The IMRT plans for thirty patients with NPC who had received radiotherapy were optimized using the VB (Plan VB) and Halcyon (Plan H) accelerators. Quality assurance verification was then conducted. The dose coverage of the planning target volume (PTV) and organs at risk (OARs), monitor units (MUs), and delivery time were analyzed for each plan. Results: All PTV and OAR indexes of Plan H and Plan VB met the clinical requirements. In the exposure dose of bilateral optic nerves between Plan H and Plan VB, no difference was found. The maximum dose of the lens, brainstem, spinal cord were 1.13 Gy, 1.36 Gy, 1.35 Gy, 2.82 Gy lower than the plan using VB , and the mean dose of the parotid glands were 3.82 Gy, 5.56 Gy lower than the plan using VB respectively, and an insignificant difference was found in the brainstem (P > 0.05). The MU for Plan H (22.92 ± 1.58 Gy) was higher than that for Plan VB (19.69 ± 4.52 Gy), and the difference was significant (P < 0.05). Conclusions: The treatment plans designed by Halcyon can meet clinical requirements with better protection for OARs and show advantages over VB in the dosimetry of NPC IMRT plans.