Prediction of three-dimensional dose distribution in intensity-modulated radiation therapy based on neural network learning / 南方医科大学学报

<p><b>OBJECTIVE</b>To establish the association between the geometric anatomical characteristics of the patients and the corresponding three-dimensional (3D) dose distribution of radiotherapy plan via feed-forward back-propagation neural network for clinical prediction of the plan dosimetric features.</p><p><b>METHODS</b>A total of 25 fixed 13-field clinical prostate cancer intensity-modulated radiation therapy (IMRT)/stereotactic body radiation therapy (SBRT) plans were collected with a prescribed dose of 50 Gy. With the distance from each voxel to the planned target volume (PTV) boundary, the distance from each voxel to each organ-at-risk (OAR), and the volume of PTV as the geometric anatomical characteristics of the patients, the voxel deposition dose was used as the plan dosimetric feature. A neural network was used to construct the correlation model between the selected input features and output dose distribution, and the model was trained with 20 randomly selected cases and verified in 5 cases.</p><p><b>RESULTS</b>The constructed model showed a small model training error, small dose differences among the verification samples, and produced accurate prediction results. In the model training, the point-to-point mean dose difference (hereinafter dose difference) of the 3D dose distribution was no greater than 0.0919∓3.6726 Gy, and the average of the relative volume values corresponding to the fixed dose sequence in the DVH (hereinafter DVH difference) did not exceed 1.7%. The dose differences among the 5 samples for validation was 0.1634∓10.5246 Gy with percent dose differences within 2.5% and DVH differences within 3%. The 3D dose distribution showed that the dose difference was small with reasonable predicted dose distribution. This model showed better performances for dose distribution prediction for bladder and rectum than for the femoral heads.</p><p><b>CONCLUSION</b>We established the relationships between the geometric anatomical characteristics of the patients and the corresponding planning 3D dose distribution via feed-forward back-propagation neural network in patients receiving IMRT/SBRT for the same tumor site. The proposed model provides individualized quality standards for automatic plan quality control.</p>

Constraint priority list-based multi-objective optimization for intensity-modulated radiation therapy / 南方医科大学学报

In intensity-modulated radiation therapy (IMRT), it is time-consuming to repeatedly adjust the objectives manually to obtain the best tradeoff between the prescribed dose of the planning target volume and sparing the organs-at-risk. Here we propose a new method to realize automatic multi-objective IMRT optimization, which quantifies the clinical preferences into the constraint priority list and adjusts the dose constraints based on the list to obtain the optimal solutions under the dose constraints. This method contains automatic adjustment mechanism of the dose constraint and automatic voxel weighting factor-based FMO model. Every time the dose constraint is adjusted, the voxel weighting factor-based FMO model is launched to find a global optimal solution that satisfied the current constraints. We tested the feasibility and effectiveness of this method in 6 cases of cervical cancer with IMRT by comparing the original plan and the automatic optimization plan generated by this method. The results showed that with the same PTV coverage and uniformity, the automatic optimization plan had a better a dose sparing of the organs-at-risk and a better plan quality than the original plan, and resulted in obvious reductions of the average V45 of the rectum from (41.99∓13.31)% to (32.55∓22.27)% and of the bladder from (44.37∓4.08)% to (28.99∓15.25)%.

Prediction of respiratory motion based on nonparametric regression for real-time tumor-tracking radiotherapy / 南方医科大学学报

<p><b>OBJECTIVE</b>It is necessary to compensate the system latencies in real-time tumor-tracking radiotherapy by prediction. However, due to the irregularities of respiratory motions, the results obtained with traditional methods were not acceptable. The purpose of this study is to evaluate the value of nonparametric regression model in respiratory motion prediction.</p><p><b>METHODS</b>The data of respiratory trajectory of 11 volunteers were obtained and predicted based on nonparametric regression method. The results were compared with those of autoregressive model and back propagation neural network. An improved method was proposed to deal with the abnormal state in respiration. We combined the prediction method with the tracking system to test its performance in practical application.</p><p><b>RESULTS</b>The results indicated that the proposed method could predict the motion accurately in real-time for different latencies. This method decreased the error of the abnormal state substantially and also allowed effective prediction of respiration motion when combined with the tracking system.</p><p><b>CONCLUSION</b>The nonparametric regression model can predict the respiratory motion accurately in real-time and therefore meets the requirement of real-time tumor-tracking radiotherapy.</p>

Monitoring of anesthetic effect based on auditory response time / 南方医科大学学报

<p><b>OBJECTIVE</b>To investigate the changes of auditory parameters during anesthesia and establish the assessment indicators for anesthesia monitoring in animal experiments.</p><p><b>METHODS</b>BALB/c mice of 4 to 6 weeks were given a single intraperitoneal dose of urethane, and the auditory evoked potential in the surgically exposed inferior colliculus in response to pure tone stimulation was recorded during urethane metabolism. The latency and amplitude data of the waves were extracted using Matlab software to analyze their variations during urethane metabolism.</p><p><b>RESULTS</b>The latency of the auditory evoked potential showed slight variation and was well correlated to time. The latency decreased progressively during urethane metabolism, fast in the initial 2 h and tending to stabilize afterwards.</p><p><b>CONCLUSION</b>The latency of the auditory evoked potential can be more suitable indicators than the amplitude for anesthetic effect monitoring.</p>

Response properties of cochlear nucleus neurons: a digital model-based study / 南方医科大学学报

<p><b>OBJECTIVE</b>To investigate the mechanisms of different response patterns of the cochlear nucleus neurons based on a digital model.</p><p><b>METHODS</b>Based on the platform of Matlab7.0 and the integrate-and-fire model, we constructed the temporal coding model.</p><p><b>RESULTS</b>The model well simulated the response types of the primary-like, chopper, onset and buildup neurons by changing the settings of some parameters related to the properties of the neurons.</p><p><b>CONCLUSIONS</b>The properties of the neurons dictate the response patterns of the cochlear nucleus neurons. The simulation results are consistent with the literature. Understanding of the response properties of the cochlear nucleus neurons can provide assistance in the study of the auditory system.</p>

Validation of an improved Demons deformable registration algorithm and its application in re-contouring in 4D-CT / 南方医科大学学报

<p><b>OBJECTIVE</b>To validate the efficiency and accuracy of an improved Demons deformable registration algorithm and evaluate its application in contour recontouring in 4D-CT.</p><p><b>METHODS</b>To increase the additional Demons force and reallocate the bilateral forces to accelerate convergent speed, we propose a novel energy function as the similarity measure, and utilize a BFGS method for optimization to avoid specifying the numbers of iteration. Mathematical transformed deformable CT images and home-made deformable phantom were used to validate the accuracy of the improved algorithm, and its effectiveness for contour recontouring was tested.</p><p><b>RESULTS</b>The improved algorithm showed a relatively high registration accuracy and speed when compared with the classic Demons algorithm and optical flow based method. Visual inspection of the positions and shapes of the deformed contours agreed well with the physician-drawn contours.</p><p><b>CONCLUSION</b>Deformable registration is a key technique in 4D-CT, and this improved Demons algorithm for contour recontouring can significantly reduce the workload of the physicians. The registration accuracy of this method proves to be sufficient for clinical needs.</p>

Cone-beam CT reconstruction based on circular scan trajectory and ring artifact reduction / 南方医科大学学报

The FDK algorithm is one of the most widely used algorithms for cone-beam CT reconstruction in circular trajectory due to its simplicity of implementation and computational efficiency. However, the images reconstructed by the FDK algorithm based on real projection data may be blurred without electronic correction and geometric calibration, and are often plagued by deleterious ring artifacts and shading artifacts. We compared the images reconstructed with and without detector correction based on computer experiment of the real biological object. The experimental simulation shows that these algorithms are effective in reducing the ring artifact without compromising the image resolution, and produce satisfactory results.

A light field-based fast computation of digitally reconstructed radiographs / 南方医科大学学报

Generation of digitally reconstructed radiographs (DDRs) is key to image-guided radiation therapy and widely used for 2D-3D image registration to verify the patient's position. The classic algorithms to generate DRRs with ray casting is time consuming to hamper the registration process. The authors describe the attempt to extend the light field rendering techniques from computer graphics, using the light field computed from ray casting to interpolate a DRRs image from a novel fixed viewpoint. Because most of the computation is performed in a preprocessing step, the speed is greatly accelerated and satisfactory DRRs can be obtained.

Genetic algorithm-based dose optimization in intensity modulated radiation therapy / 南方医科大学学报

As a powerful global optimization approach, genetic algorithms (GA) can solve a variety of optimization problems in which the objective function is discontinuous, non-differentiable, or highly non-linear, to produce high convergence speed and vast search space. In this thesis, GA is used to optimize the beam weights of intensity modulated radiation therapy (IMRT) inverse planning, and 2D and 3D isodose contour as well as dose volume histogram (DVH) are used to evaluate the treatment plan. Also presented in this thesis are the results of calculation with discussions.

A preliminary study of beam weight optimization of intensity-modulated radiation therapy with genetic algorithm / 南方医科大学学报

<p><b>UNLABELLED</b>To study the method for dose calculation and beam weight optimization of intensity-modulated radiation therapy (IMRT).</p><p><b>METHODS</b>The IMRT dose calculation model based on two-dimensional convolution was constructed, the program of dose calculation and beam weight optimization with genetic algorithm was written with Visual c#.Net, and the optimization results were analyzed.</p><p><b>RESULTS</b>Genetic algorithm optimization of beam weights can produce highly conformal dose distributions within a clinically acceptable computation time.</p><p><b>CONCLUSION</b>Genetic algorithm is valid and efficient in IMRT beam weight optimization, which may facilitate IMRT treatment planning.</p>