Convolutional neural network enhancement of fast-scan low-dose cone-beam CT images for head and neck radiotherapy.
Convolutional neural network enhancement of fast-scan low-dose cone-beam CT images for head and neck radiotherapy.
复制标题
头颈放疗快速扫描低剂量锥束CT图像的卷积神经网络增强。
DOI：
10.1088/1361-6560/ab6240
复制
发表时间：
2020-01-27
影响因子：
3.5
通讯作者：
Rong Y
中科院分区：
文献类型：
--
作者：
Yuan N;Dyer B;Rao S;Chen Q;Benedict S;Shang L;Kang Y;Qi J;Rong Y
通过深度学习卷积神经网络（CNN）方法提高头颈部（HN）放疗快速扫描低剂量锥形束计算机断层扫描（CBCT）的图像质量和CT数精度。回顾性分析55例HN患者的配对CBCT和CT图像。其中15例患者在治疗期间进行了适应性重新规划，即当天CT/CBCT对。其余40例患者（术后）有配对的计划CT和一阶CBCT图像，解剖改变很小。CNN选择了5个深度27层的2D U-Net架构。CNN训练使用来自40例术后HN患者的2080对CT/CBCT切片数据。验证和测试数据集来自接受适应性重新规划的患者，分别包括5个当日数据集（260对切片）和10个当日数据集（520对切片）。为了检验训练数据集选择和网络性能差异作为训练数据大小的函数的影响，使用30、40和50个数据集训练额外的网络。采用霍斯菲尔德单位（Hounsfield units， HU）的平均绝对误差（MAE）、信噪比（SNR）和结构相似度（SSIM）对增强后的CBCT图像质量与CT图像进行定量比较。增强的CBCT图像减少了伪影失真，提高了软组织对比度。使用40个数据集训练的网络的成像性能与使用50个数据集训练的网络相当，并且优于使用30个数据集训练的网络。CBCT和增强CBCT图像的比较表明，平均MAE从172.73提高到49.28 HU，信噪比从8.27提高到14.25 dB， SSIM从0.42提高到0.85。使用NVIDIA GeForce GTX 1080 Ti GPU，每位患者的图像处理时间为2秒。所提出的深度学习方法对于快速扫描低剂量CBCT的图像质量增强既快速又有效。该方法具有支持HN癌症患者快速在线适应性重新规划的潜力。
To improve image quality and CT number accuracy of fast-scan low-dose cone-beam computed tomography (CBCT) through a deep-learning convolutional neural network (CNN) methodology for head-and-neck (HN) radiotherapy. Fifty-five paired CBCT and CT images from HN patients were retrospectively analysed. Among them, 15 patients underwent adaptive replanning during treatment, thus had same-day CT/CBCT pairs. The remaining 40 patients (post-operative) had paired planning CT and 1st fraction CBCT images with minimal anatomic changes. A 2D U-Net architecture with 27-layers in 5 depths was chosen for the CNN. CNN training was performed using data from 40 post-operative HN patients with 2080 paired CT/CBCT slices. Validation and test datasets were from patients undergoing adaptive replanning and include 5 same-day datasets with 260 slice pairs and 10 same-day datasets with 520 slice pairs, respectively. To examine the impact of differences in training dataset selection and network performance as a function of training data size, additional networks were trained using 30, 40 and 50 datasets. Image quality of enhanced CBCT images were quantitatively compared against the CT image using mean absolute error (MAE) of Hounsfield units (HU), signal-to-noise ratio (SNR) and structural similarity (SSIM). Enhanced CBCT images reduced artifact distortion and improved soft tissue contrast. Networks trained with 40 datasets had imaging performance comparable to those trained with 50 datasets and outperformed those trained with 30 datasets. Comparison of CBCT and enhanced CBCT images demonstrated improvement in average MAE from 172.73 to 49.28 HU, SNR from 8.27 to 14.25 dB, and SSIM from 0.42 to 0.85. The image processing time is 2 seconds per patient using a NVIDIA GeForce GTX 1080 Ti GPU. The proposed deep-leaning methodology was both fast and effective for image quality enhancement of fast-scan low-dose CBCT. This method has potential to support fast online-adaptive re-planning for HN cancer patients.
登录
查看更多内容
DOI：
10.1016/j.radonc.2011.05.028
发表时间：
2011-12
期刊：
Radiotherapy and oncology : journal of the European Society for Therapeutic Radiology and Oncology
影响因子：
--
作者：
Simone CB 2nd;Ly D;Dan TD;Ondos J;Ning H;Belard A;O'Connell J;Miller RW;Simone NL
通讯作者：
Simone NL
影响因子：
3.5
作者：
Tian Z;Jia X;Yuan K;Pan T;Jiang SB
通讯作者：
Jiang SB
DOI：
10.1016/s0360-3016(02)02884-5
发表时间：
2002-08-01
影响因子：
7
作者：
Jaffray, DA;Siewerdsen, JH;Martinez, AA
通讯作者：
Martinez, AA
DOI：
10.7759/cureus.2548
发表时间：
2018-04-29
期刊：
Cureus
影响因子：
--
作者：
Kida S;Nakamoto T;Nakano M;Nawa K;Haga A;Kotoku J;Yamashita H;Nakagawa K
通讯作者：
Nakagawa K
影响因子：
3.5
作者：
Jia, Xun;Dong, Bin;Jiang, Steve B.
通讯作者：
Jiang, Steve B.