Payam Samadi Miandoab, Yaoying Liu, Xuying Shang, Tie Lv, Hui jun Xu, Gaolong Zhang, Shouping Xu
{"title":"CNN-GRU-Dense模型用于放疗期间肝脏肿瘤实时跟踪的可行性研究","authors":"Payam Samadi Miandoab, Yaoying Liu, Xuying Shang, Tie Lv, Hui jun Xu, Gaolong Zhang, Shouping Xu","doi":"10.1002/mp.70014","DOIUrl":null,"url":null,"abstract":"<div>\n \n \n <section>\n \n <h3> Background</h3>\n \n <p>Real-time tumor tracking helps overcome challenges in delivering accurate radiotherapy. Commercial tracking devices use a hybrid external–internal correlation model (ECM) that combines intermittent X-ray imaging of the tumor's internal position with continuous monitoring of external respiratory motion. This approach improves tracking accuracy and treatment effectiveness.</p>\n </section>\n \n <section>\n \n <h3> Purpose</h3>\n \n <p>This study simulated using a deep learning model (CNN-GRU-Dense model) to track liver tumors in real-time during treatment—without needing ongoing updates. The model's accuracy was tested against several well-known methods, including the hybrid correlation model used in the CyberKnife system, the NG-RC model, and the augmented linear model.</p>\n </section>\n \n <section>\n \n <h3> Methods</h3>\n \n <p>The CNN-GRU-Dense model comprises convolutional, Gated Recurrent Units (GRU), and dense layers to estimate tumor position in various directions. Initially, input signals are processed through a 1D convolutional layer that employs 64 filters with a kernel size of 3 and ReLU activation to extract spatial features. Next, the extracted features are processed by two stacked GRU layers, each containing 256 units with ReLU activation, enabling the model to capture temporal dependencies. After the GRU layers, the data undergoes refinement through two dense (fully connected) layers, each with 64 units and ReLU activation, ensuring enhanced feature extraction. Finally, the output is passed through a single-unit output layer with linear activation, providing the estimated tumor position. For training the CNN-GRU-Dense model, 26 min of motion patterns (a patient-specific data) are utilized. The proposed model underwent hyperparameter optimization using the RandomSearch approach. This method explored a broad search space, which included the number of filters and kernel size in the 1D Convolutional layer, the number of GRU units, the number of fully connected dense layers, the learning rate, and the loss function. Using a learning rate 0.001, the model was optimized with the Adam optimizer and trained with the mean squared error (MSE) loss function. The training was conducted for 30 epochs with a batch size of 300, aiming to strike a balance between speed and stability during the learning process. Finally, the trained CNN-GRU-Dense model was tested with new external motion data to estimate tumor positions. The model parameters remain unchanged throughout the treatment, requiring no updates. Fifty-seven motion trace datasets from the CyberKnife system were used to evaluate the CNN-GRU-Dense model performance. These traces were grouped into three liver regions: central, lower, and upper.</p>\n </section>\n \n <section>\n \n <h3> Results</h3>\n \n <p>The CNN-GRU-Dense model demonstrated improved estimation accuracy compared to other ECMs (Wilcoxon signed rank <i>p </i>< 0.05). The 3D radial estimation accuracy (Mean ± standard deviations (SD)) using the CyberKnife system, the NG-RC model, the augmented linear model, and the CNN-GRU-Dense model was 1.42 ± 0.44 mm, 1.23 ± 0.75 mm, 0.71 ± 0.40 mm, and 0.55 ± 0.27 mm, respectively.</p>\n </section>\n \n <section>\n \n <h3> Conclusions</h3>\n \n <p>The simulation results showed that the CNN-GRU-Dense model outperformed several existing methods, including the augmented linear model used in standard linear accelerators, the NG-RC model, and the constrained fourth-order polynomial equations used in the CyberKnife and Radixact systems. One key advantage of the CNN-GRU-Dense model is that it doesn't need to be updated during treatment, which reduces patients' radiation exposure.</p>\n </section>\n </div>","PeriodicalId":18384,"journal":{"name":"Medical physics","volume":"52 10","pages":""},"PeriodicalIF":3.2000,"publicationDate":"2025-09-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Feasibility study of using CNN-GRU-Dense model for real-time liver tumor tracking during radiotherapy\",\"authors\":\"Payam Samadi Miandoab, Yaoying Liu, Xuying Shang, Tie Lv, Hui jun Xu, Gaolong Zhang, Shouping Xu\",\"doi\":\"10.1002/mp.70014\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<div>\\n \\n \\n <section>\\n \\n <h3> Background</h3>\\n \\n <p>Real-time tumor tracking helps overcome challenges in delivering accurate radiotherapy. Commercial tracking devices use a hybrid external–internal correlation model (ECM) that combines intermittent X-ray imaging of the tumor's internal position with continuous monitoring of external respiratory motion. This approach improves tracking accuracy and treatment effectiveness.</p>\\n </section>\\n \\n <section>\\n \\n <h3> Purpose</h3>\\n \\n <p>This study simulated using a deep learning model (CNN-GRU-Dense model) to track liver tumors in real-time during treatment—without needing ongoing updates. The model's accuracy was tested against several well-known methods, including the hybrid correlation model used in the CyberKnife system, the NG-RC model, and the augmented linear model.</p>\\n </section>\\n \\n <section>\\n \\n <h3> Methods</h3>\\n \\n <p>The CNN-GRU-Dense model comprises convolutional, Gated Recurrent Units (GRU), and dense layers to estimate tumor position in various directions. Initially, input signals are processed through a 1D convolutional layer that employs 64 filters with a kernel size of 3 and ReLU activation to extract spatial features. Next, the extracted features are processed by two stacked GRU layers, each containing 256 units with ReLU activation, enabling the model to capture temporal dependencies. After the GRU layers, the data undergoes refinement through two dense (fully connected) layers, each with 64 units and ReLU activation, ensuring enhanced feature extraction. Finally, the output is passed through a single-unit output layer with linear activation, providing the estimated tumor position. For training the CNN-GRU-Dense model, 26 min of motion patterns (a patient-specific data) are utilized. The proposed model underwent hyperparameter optimization using the RandomSearch approach. This method explored a broad search space, which included the number of filters and kernel size in the 1D Convolutional layer, the number of GRU units, the number of fully connected dense layers, the learning rate, and the loss function. Using a learning rate 0.001, the model was optimized with the Adam optimizer and trained with the mean squared error (MSE) loss function. The training was conducted for 30 epochs with a batch size of 300, aiming to strike a balance between speed and stability during the learning process. Finally, the trained CNN-GRU-Dense model was tested with new external motion data to estimate tumor positions. The model parameters remain unchanged throughout the treatment, requiring no updates. Fifty-seven motion trace datasets from the CyberKnife system were used to evaluate the CNN-GRU-Dense model performance. These traces were grouped into three liver regions: central, lower, and upper.</p>\\n </section>\\n \\n <section>\\n \\n <h3> Results</h3>\\n \\n <p>The CNN-GRU-Dense model demonstrated improved estimation accuracy compared to other ECMs (Wilcoxon signed rank <i>p </i>< 0.05). The 3D radial estimation accuracy (Mean ± standard deviations (SD)) using the CyberKnife system, the NG-RC model, the augmented linear model, and the CNN-GRU-Dense model was 1.42 ± 0.44 mm, 1.23 ± 0.75 mm, 0.71 ± 0.40 mm, and 0.55 ± 0.27 mm, respectively.</p>\\n </section>\\n \\n <section>\\n \\n <h3> Conclusions</h3>\\n \\n <p>The simulation results showed that the CNN-GRU-Dense model outperformed several existing methods, including the augmented linear model used in standard linear accelerators, the NG-RC model, and the constrained fourth-order polynomial equations used in the CyberKnife and Radixact systems. One key advantage of the CNN-GRU-Dense model is that it doesn't need to be updated during treatment, which reduces patients' radiation exposure.</p>\\n </section>\\n </div>\",\"PeriodicalId\":18384,\"journal\":{\"name\":\"Medical physics\",\"volume\":\"52 10\",\"pages\":\"\"},\"PeriodicalIF\":3.2000,\"publicationDate\":\"2025-09-21\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Medical physics\",\"FirstCategoryId\":\"3\",\"ListUrlMain\":\"https://aapm.onlinelibrary.wiley.com/doi/10.1002/mp.70014\",\"RegionNum\":2,\"RegionCategory\":\"医学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q1\",\"JCRName\":\"RADIOLOGY, NUCLEAR MEDICINE & MEDICAL IMAGING\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Medical physics","FirstCategoryId":"3","ListUrlMain":"https://aapm.onlinelibrary.wiley.com/doi/10.1002/mp.70014","RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"RADIOLOGY, NUCLEAR MEDICINE & MEDICAL IMAGING","Score":null,"Total":0}
引用次数: 0
摘要
实时肿瘤跟踪有助于克服提供准确放疗的挑战。商业跟踪设备使用混合的内外相关模型(ECM),将肿瘤内部位置的间歇性x射线成像与外部呼吸运动的连续监测相结合。这种方法提高了跟踪精度和治疗效果。本研究使用深度学习模型(CNN-GRU-Dense模型)进行模拟,在治疗过程中实时跟踪肝脏肿瘤,而无需持续更新。该模型的准确性通过几种知名方法进行了测试,包括射波刀系统中使用的混合相关模型、NG-RC模型和增强线性模型。方法CNN-GRU-Dense模型由卷积层、门控循环单元(GRU)层和密集层组成,在各个方向上估计肿瘤的位置。最初,输入信号通过一个1D卷积层进行处理,该层采用64个滤波器,核大小为3,并使用ReLU激活来提取空间特征。接下来,提取的特征由两个堆叠的GRU层处理,每个GRU层包含256个具有ReLU激活的单元,使模型能够捕获时间依赖性。在GRU层之后,数据通过两个密集(完全连接)层进行细化,每个层有64个单元和ReLU激活,确保增强的特征提取。最后,输出通过线性激活的单单元输出层,提供估计的肿瘤位置。为了训练CNN-GRU-Dense模型,使用了26分钟的运动模式(特定于患者的数据)。采用随机搜索方法对模型进行超参数优化。该方法探索了一个广泛的搜索空间,包括1D卷积层中滤波器的数量和核的大小、GRU单元的数量、完全连接的密集层的数量、学习率和损失函数。使用学习率0.001,使用Adam优化器对模型进行优化,并使用均方误差(MSE)损失函数对模型进行训练。训练进行了30个epoch, batch大小为300,目的是在学习过程中达到速度和稳定性的平衡。最后,使用新的外部运动数据对训练好的CNN-GRU-Dense模型进行测试,以估计肿瘤位置。模型参数在整个治疗过程中保持不变,不需要更新。使用来自射波刀系统的57个运动跟踪数据集来评估CNN-GRU-Dense模型的性能。这些痕迹被分为三个肝脏区域:中央、下、上。结果与其他ecm相比,CNN-GRU-Dense模型显示出更高的估计精度(Wilcoxon signed rank p < 0.05)。射波刀系统、NG-RC模型、增强线性模型和CNN-GRU-Dense模型的三维径向估计精度(Mean±standard deviation (SD))分别为1.42±0.44 mm、1.23±0.75 mm、0.71±0.40 mm和0.55±0.27 mm。仿真结果表明,CNN-GRU-Dense模型优于现有的几种方法,包括用于标准线性加速器的增强线性模型、NG-RC模型以及用于射波刀和Radixact系统的约束四阶多项式方程。CNN-GRU-Dense模型的一个关键优势是它不需要在治疗期间更新,从而减少了患者的辐射暴露。
Feasibility study of using CNN-GRU-Dense model for real-time liver tumor tracking during radiotherapy
Background
Real-time tumor tracking helps overcome challenges in delivering accurate radiotherapy. Commercial tracking devices use a hybrid external–internal correlation model (ECM) that combines intermittent X-ray imaging of the tumor's internal position with continuous monitoring of external respiratory motion. This approach improves tracking accuracy and treatment effectiveness.
Purpose
This study simulated using a deep learning model (CNN-GRU-Dense model) to track liver tumors in real-time during treatment—without needing ongoing updates. The model's accuracy was tested against several well-known methods, including the hybrid correlation model used in the CyberKnife system, the NG-RC model, and the augmented linear model.
Methods
The CNN-GRU-Dense model comprises convolutional, Gated Recurrent Units (GRU), and dense layers to estimate tumor position in various directions. Initially, input signals are processed through a 1D convolutional layer that employs 64 filters with a kernel size of 3 and ReLU activation to extract spatial features. Next, the extracted features are processed by two stacked GRU layers, each containing 256 units with ReLU activation, enabling the model to capture temporal dependencies. After the GRU layers, the data undergoes refinement through two dense (fully connected) layers, each with 64 units and ReLU activation, ensuring enhanced feature extraction. Finally, the output is passed through a single-unit output layer with linear activation, providing the estimated tumor position. For training the CNN-GRU-Dense model, 26 min of motion patterns (a patient-specific data) are utilized. The proposed model underwent hyperparameter optimization using the RandomSearch approach. This method explored a broad search space, which included the number of filters and kernel size in the 1D Convolutional layer, the number of GRU units, the number of fully connected dense layers, the learning rate, and the loss function. Using a learning rate 0.001, the model was optimized with the Adam optimizer and trained with the mean squared error (MSE) loss function. The training was conducted for 30 epochs with a batch size of 300, aiming to strike a balance between speed and stability during the learning process. Finally, the trained CNN-GRU-Dense model was tested with new external motion data to estimate tumor positions. The model parameters remain unchanged throughout the treatment, requiring no updates. Fifty-seven motion trace datasets from the CyberKnife system were used to evaluate the CNN-GRU-Dense model performance. These traces were grouped into three liver regions: central, lower, and upper.
Results
The CNN-GRU-Dense model demonstrated improved estimation accuracy compared to other ECMs (Wilcoxon signed rank p < 0.05). The 3D radial estimation accuracy (Mean ± standard deviations (SD)) using the CyberKnife system, the NG-RC model, the augmented linear model, and the CNN-GRU-Dense model was 1.42 ± 0.44 mm, 1.23 ± 0.75 mm, 0.71 ± 0.40 mm, and 0.55 ± 0.27 mm, respectively.
Conclusions
The simulation results showed that the CNN-GRU-Dense model outperformed several existing methods, including the augmented linear model used in standard linear accelerators, the NG-RC model, and the constrained fourth-order polynomial equations used in the CyberKnife and Radixact systems. One key advantage of the CNN-GRU-Dense model is that it doesn't need to be updated during treatment, which reduces patients' radiation exposure.
期刊介绍:
Medical Physics publishes original, high impact physics, imaging science, and engineering research that advances patient diagnosis and therapy through contributions in 1) Basic science developments with high potential for clinical translation 2) Clinical applications of cutting edge engineering and physics innovations 3) Broadly applicable and innovative clinical physics developments
Medical Physics is a journal of global scope and reach. By publishing in Medical Physics your research will reach an international, multidisciplinary audience including practicing medical physicists as well as physics- and engineering based translational scientists. We work closely with authors of promising articles to improve their quality.