Purpose Clinical target volumes (CTV) and organs at risk (OAR) could be auto-contoured to save workload. The goal of this study was to assess a convolutional neural network (CNN) for totally automatic and accurate CTV and OAR in prostate cancer, while also comparing anticipated treatment plans based on auto-contouring CTV to clinical plans. Methods From January 2013 to January 2019, 217 computed tomography (CT) scans of patients with locally advanced prostate cancer treated at our hospital were collected and analyzed. CTV and OAR were delineated with a deep learning based method, which named CUNet. The performance of this strategy was evaluated using the mean Dice similarity coefficient (DSC), 95th percentile Hausdorff distance (95HD), and subjective evaluation. Treatment plans were graded using predetermined evaluation criteria, and % errors for clinical doses to the planned target volume (PTV) and organs at risk(OARs) were calculated. Results The defined CTVs had mean DSC and 95HD values of 0.84 and 5.04 mm, respectively. For one patient's CT scans, the average delineation time was less than 15 seconds. When CTV outlines from CUNetwere blindly chosen and compared to GT, the overall positive rate in clinicians A and B was 53.15% vs 46.85%, and 54.05% vs 45.95%, respectively (P>0.05), demonstrating that our deep machine learning model performed as good as or better than human demarcation Furthermore, 8 testing patients were chosen at random to design the predicted plan based on the auto-courtoring CTV and OAR, demonstrating acceptable agreement with the clinical plan: average absolute dose differences of D2, D50, D98, Dmean for PTV are within 0.74%, and average absolute volume differences of V45, V50 for OARs are within 3.4%. Without statistical significance (p>0.05), the projected findings are comparable to clinical truth. Conclusion The experimental results show that the CTV and OARs defined by CUNet for prostate cancer were quite close to the ground reality.CUNet has the potential to cut radiation oncologists' contouring time in half. When compared to clinical plans, the differences between estimated doses to CTV and OAR based on auto-courtoring were small, with no statistical significance, indicating that treatment planning for prostate cancer based on auto-courtoring has potential.

Background and Objective: Delineation of the clinical target volume (CTV) and organs at risk (OARs) is very important for radiotherapy but is time-consuming and prone to inter- and intra-observer variation. We trained and evaluated a U-Net-based model to provide fast and consistent auto-segmentation for breast cancer radiotherapy. Methods: We collected 160 patients’ computed tomography (CT) scans with early-stage breast cancer who underwent breast-conserving surgery (BCS) and were treated with radiotherapy in our center. CTV and OARs (contralateral breast, heart, lungs and spinal cord) were delineated manually by two experienced radiation oncologists. The data were used for model training and testing. The dice similarity coefficient (DSC) and 95th Hausdorff distance (95HD) were used to assess the performance of our model. CTV and OARs were randomly selected as ground truth (GT) masks, and artificial intelligence (AI) masks were generated by the proposed model. The contours were randomly distributed to two clinicians to compare CTV score differences. The consistency between two clinicians was tested. We also evaluated time cost for auto-delineation. Results: The mean DSC values of the proposed method were 0.94, 0.95, 0.94, 0.96, 0.96 and 0.93 for breast CTV, contralateral breast, heart, right lung, left lung and spinal cord, respectively. The mean 95HD values were 4.31 mm, 3.59 mm, 4.86 mm, 3.18 mm, 2.79 mm and 4.37 mm for the above structures respectively. The average CTV scores for AI and GT were 2.92 versus 2.89 when evaluated by oncologist A (P=.612), and 2.75 versus 2.83 by oncologist B (P=.213), with no statistically significant differences. The consistency between two clinicians was poor (Kappa=0.282). The times for auto-segmentation of CTV and OARs were 3.88 s and 6.15 s. Conclusions: Our proposed model can improve the speed and accuracy of delineation compared with U-Net, while it performed equally well with the segmentation generated by oncologists.