Search

Background Prostate alignment is subject to interobserver variability in cone-beam CT (CBCT)-based soft-tissue matching. This study aims to analyze the impact of possible interobserver variability in CBCT-based soft-tissue matching for prostate cancer radiotherapy. Methods Retrospective data, consisting of 156 CBCT images from twelve prostate cancer patients with elective nodal irradiation were analyzed in this study. To simulate possible interobserver variability, couch shifts of 2 mm relative to the resulting patient position of prostate alignment were assumed as potential patient positions (27 possibilities). For each CBCT, the doses of the potential patient positions were re-calculated using deformable image registration-based synthetic CT. The impact of the simulated interobserver variability was evaluated using tumor control probabilities (TCPs) and normal tissue complication probabilities (NTCPs). Results No significant differences in TCPs were found between prostate alignment and potential patient positions (0.944 ± 0.003 vs 0.945 ± 0.003, P = 0.117). The average NTCPs of the rectum ranged from 5.16 to 7.29 (%) among the potential patient positions and were highly influenced by the couch shift in the anterior–posterior direction. In contrast, the average NTCPs of the bladder ranged from 0.75 to 1.12 (%) among the potential patient positions and were relatively negligible. Conclusions The NTCPs of the rectum, rather than the TCPs of the target, were highly influenced by the interobserver variability in CBCT-based soft-tissue matching. This study provides a theoretical explanation for daily CBCT-based image guidance and the prostate-rectum interface matching procedure. Trial registration: Not applicable.

Background Surface-guided radiation therapy can be used to continuously monitor a patient’s surface motions during radiotherapy by a non-irradiating, noninvasive optical surface imaging technique. In this study, machine learning methods were applied to predict external respiratory motion signals and predict internal liver motion in this therapeutic context. Methods Seven groups of interrelated external/internal respiratory liver motion samples lasting from 5 to 6 min collected simultaneously were used as a dataset, Dv. Long short-term memory (LSTM) and support vector regression (SVR) networks were then used to establish external respiratory signal prediction models (LSTMpred/SVRpred) and external/internal respiratory motion correlation models (LSTMcorr/SVRcorr). These external prediction and external/internal correlation models were then combined into an integrated model. Finally, the LSTMcorr model was used to perform five groups of model updating experiments to confirm the necessity of continuously updating the external/internal correlation model. The root-mean-square error (RMSE), mean absolute error (MAE), and maximum absolute error (MAX_AE) were used to evaluate the performance of each model. Results The models established using the LSTM neural network performed better than those established using the SVR network in the tasks of predicting external respiratory signals for latency-compensation (RMSE Conclusions The LSTM networks outperform SVR networks at predicting external respiratory signals and internal liver motion because of LSTM’s strong ability to deal with time-dependencies. The LSTM-based integrated model performs well at predicting liver motion from external respiratory signals with system latencies of up to 450 ms. It is necessary to update the external/internal correlation model continuously.

Background Surface-guided radiation therapy (SGRT) employs a non-invasive real-time optical surface imaging (OSI) technique for patient surface motion monitoring during radiotherapy. The main purpose of this study is to verify the real-time tracking accuracy of SGRT for respiratory motion and provide a fitting method to detect the time delay of gating. Methods A respiratory motion phantom was utilized to simulate respiratory motion using 17 cosine breathing pattern curves with various periods and amplitudes. The motion tracking of the phantom was performed by the Catalyst™ system. The tracking accuracy of the system (with period and amplitude variations) was evaluated by analyzing the adjusted coefficient of determination (A_R2) and root mean square error (RMSE). Furthermore, 13 actual respiratory curves, which were categorized into regular and irregular patterns, were selected and then simulated by the phantom. The Fourier transform was applied to the respiratory curves, and tracking accuracy was compared through the quantitative analyses of curve similarity using the Pearson correlation coefficient (PCC). In addition, the time delay of amplitude-based respiratory-gating radiotherapy based on the OSI system with various beam hold times was tested using film dosimetry for the Elekta Versa-HD and Varian Edge linacs. A dose convolution-fitting method was provided to accurately measure the beam-on and beam-off time delays. Results A_R2 and RMSE for the cosine curves were 0.9990–0.9996 and 0.110–0.241 mm for periods ranging from 1 s to 10 s and 0.9990–0.9994 and 0.059–0.175 mm for amplitudes ranging from 3 mm to 15 mm. The PCC for the actual respiratory curves ranged from 0.9955 to 0.9994, which was not significantly affected by breathing patterns. For gating radiotherapy, the average beam-on and beam-off time delays were 1664 ± 72 and 25 ± 30 ms for Versa-HD and 303 ± 45 and 34 ± 25 ms for Edge, respectively. The time delay was relatively stable as the beam hold time increased. Conclusions The OSI technique provides high accuracy for respiratory motion tracking. The proposed dose convolution-fitting method can accurately measure the time delay of respiratory-gating radiotherapy. When the OSI technique is used for respiratory-gating radiotherapy, the time delay for the beam-on is considerably longer than the beam-off.