基于卷积神经网络CT/CBCT影像质量自动分析

目的:通过独立的程序自动分析数据,可以在减轻影像的质量保证(QA)工作量的同时,尽可能避免操作者主观因素造成的偏差。方法:对Catphan500/503/504/600的CT/CBCT影像按照功能模块进行分类,并通过卷积神经网络(CNN)进行学习,学习后对新输入的CT/CBCT影像可以自动识别并根据功能模块进行分类,继而对相关指标包括影像CT值的线性、调制传递函数以及均匀性等进行自动分析,以便确保临床应用的影像质量达到要求。结果:对于Catphan500扫描的CT图像和Catphan503扫描的CBCT图像,经过CNN自动分类对于功能模块CTP401、CTP404、CTP528都可以正确标记出来,但是CTP486的精确度没有达到100%,即有部分不属于CTP486的模块被错误判断成CTP486。同时均可实现对CT的值线性、调制传递函数以及均匀性3个图像指标进行自动分析。结论:基于CNN能够准确地对CT/CBCT扫描的Catphan图像进行分类,下一步将拓展该方法到其他影像设备的QA体模,以便实现更广泛的自动影像质量保证。     

Varian On-Board Imager(OBI)常规测量方法及质量保证

目的:通过建立一套OBI系统的常规测量方法,探讨放射治疗中利用OBI系统进行摆位验证的质量保证(QA)内容.方法:利用质量保证工具和体模对VarianIX加速器的OBI系统进行每日、每周、每月、每年的常规测量,完成质量保证项目检测.结果:OBI系统的操作安全性,几何准确性和图像质量均在允许范围内.OBI系统的等中心准确...     

OBI影像引导放射治疗系统的质量保证（QA）及维护保养

目的：利用瓦里安厂家提供的质量保证工具。对瓦里安直线加速器的机载影像引导系统进行常规测量,保证放射治疗中患者位置验证的准确性和安全性。通过对设备日常维护保养的介绍,使操作人员熟悉OBI设备的维护保养,为治疗工作的顺利进行,创造良好的设备环境。方法：根据影像引导系统OBI（OnBoardImager）的验收报告和测量方法,对OBI影像引导系统的安全性、功能性、几何准确性和图像质量进行检测,完成质量保证项目的检测及维护保养。结果：影像引导系统的安全性和功能性均正常,几何准确性方面和图像质量均在允许范围内。未出现因保养不及时而发生的故障。结论：定期的质量保证检测和维护保养表明,OBI系统性能稳定可靠,各项参数指标在允许范围内。定期的维护保养可以有效的减少故障率的发生,提高设备的安全性和可靠性。     

基于Web的剂量体积直方图数据自动提取工具开发及验证

目的:开发并测试一套基于Web的剂量体积直方图(DVH)数据自动提取工具。方法:(1)采用Django应用框架和Python编程语言,设计一套基于Web的DVH数据自动提取工具。(2)利用自动工具分析从Eclipse计划系统中导出的30例相同类型计划的DVH表单数据,并采用人工方法读取DVH图中的相应参数作为对比,分析其耗时、准确性等方面的表现及误差产生原因。结果:自动提取DVH数据的效率远高于人工分析,且正确率更有保障。对于计划靶区体积的均匀性指数、股骨头和膀胱D_(50%)和平均剂量等参数,自动与人工提取的差异极小(误差≤0.01%,P0.05)。但对于适形指数(CI)值的计算,由于计划系统空间采样算法的原因使得基于等剂量线结构转换测量以及DVH表单数据分析之间的结果存在较大差异[CI_PGTV平均相差(2.60±1.04)%,CI_PTV平均相差(0.66±0.29)%,P0.001],但Web工具采用的DVH分析结果更加接近Eclipse自动生成的CI值,且有效避免了后者一次只能计算一个CI值的缺陷。结论:本工作开发的基于Web的工具可以对海量DVH数据进行高效、准确的自动统计,且具有跨平台应用等优势。     

基于Varian锥形束CT影像质量的直线加速器选择策略

目的:定量分析不同锥形束CT（CBCT）影像特点,从而为患者选择最佳设备。方法:利用CatPhan604模体分析Edge、TrueBeam及新旧ix机载CBCT头、胸、盆模式图像。结果:12组图像头、胸、盆CT值最准确的是ix新机器、TrueBeam、Edge,分别为5.69、0.81、6.74 HU;CT值线性最好的是ix旧机器或新机器、Edge、Edge,分别为0.995、0.996、0.997;线性距离误差最小的是ix旧机器、Edge、Edge或TrueBeam或ix旧机器,分别为0.050、0.075、0.100 mm;角度误差最小的是ix旧机器、Edge或TrueBeam、Edge或ix新机器,分别为0.075°、0.050°、0.075°。头、胸、盆高对比度分辨率最好的是ix旧机器、Edge、Edge,分别为7、5、5 LP/cm;均匀性最好的是Edge、Edge、Edge,分别为4.78、20.19、4.63。头、胸、盆噪声最好的是Edge、ix新机器、ix新机器,分别为27.53、8.67、7.33;信噪比最好的是Edge、TrueBeam、ix新机器,分别为83.17、124.39、288.39;对比度噪声比最好的是Edge、ix新机器、ix新机器,分别为11.92、41.42、51.47。低对比度分辨率头部未可见,胸、盆部最好的是Edge或TrueBeam、Edge,分别为6.00、3.75。结论:CBCT系统间差异大,为患者选择加速器时应考虑成像特点,如自适应放疗选择高CT值线性和准确性,立体定向放疗选择低距离和角度误差设备等。     

电子射野影像系统在加速器辐射野测量中的应用探讨

目的:探讨电子射野影像系统(EPID)在加速器辐射野与灯光野一致性测量中的应用。方法:使用Varian 600CD医用电子直线加速器,6MV X射线能量,使用水平尺,确认机架位于0°,准直器0°,提前校准照射野中心和投影十字线,将厂家自带的金属点十字影子板插在加速器机头上,金属点十字影子板上两金点之间在SSD=100cm处的投影距离为1cm,调整机头十字线与金属十字线投影重合;打开EPID测量板,在SSD=100cm条件下,灯光野分别开到标准野(10×10)cm,(15×15)cm,(20×20)cm,(25×25)cm,剂量率100MU/M,曝光5MU;得到各标准野的辐射野,两金属点之间标准距离1cm,使用测量软件分别分别测量辐射野各方向距离。结果:辐射野各方向偏差较小,均小于±2mm。结论:EPID射野影像检测方式适合于临床质控检验,可用于加速器辐射野与灯光野一致性的质控测量,减少工作量。     

DoseLab软件检测CT图像噪声的程序改进及应用分析

目的:对DoseLab软件进行程序改进,增加检测CT图像噪声的功能,对改进的程序进行测试分析。方法:首先,通过使用圆的内接多边形顶点位置计算公式,得到圆内接正三十二边形顶点坐标值。然后,在DoseLab软件Catphan 504模体CTP486模块的图像分析程序中,添加一个正三十二边形的感兴趣区（ROI）,用于检测CT图像噪声。选取2018年每月由西门子CT模拟机日常质量检测（DQC）程序得到的水模体两个层面（S3和S4）的CT图像,对DoseLab改进程序进行测试。对DoseLab改进程序和DQC程序得到的CT图像噪声数据,进行统计分析和比较研究。结果:根据公式计算得到了半径4 cm圆的内接正三十二边形的32个顶点的坐标值,该多边形ROI的面积为49.94 cm2。计算DoseLab改进程序和DQC程序得到的CT图像噪声的差异（ΔN）。在120 kV情形,S3和S4层的ΔN值分别为（0.06±0.07） HU和（0.03±0.09） HU;在140 kV情形,S3和S4层的ΔN值分别为（0.10±0.09） HU和（0.08±0.09） HU。结论:通过添加正三十二边形ROI得到的DoseLab改进程序,可以自动分析水模体和Catphan模体,得到CT图像噪声数据。     

不同探测器在多叶准直器质量保证中的定位精度比较

目的:比较辐射自显影胶片、电子射野影像系统、电离室矩阵等不同探测器在多叶准直器质量保证中的定位精度。方法:采用辐射自显影胶片(GAFCHROMIC EBT胶片)、电子射野影像系统、电离室矩阵(IBA公司Matrixx和PTW公司Seven29)测量和比较瓦里安公司Clinac ix加速器的多叶准直器叶片的边缘的边响应函数,比较测量结果,评价不同探测器的定位精度。结果:四种探测器的定位精度均可达到0.1mm,其中电子射野影像系统的灵敏度最高。结论:上述探测器均能满足临床质控需要。     

基于人工智能的正位DR胸片质控体系研究与应用

随着医学诊断、治疗模式的改变,医学影像的质量直接影响着医生对病情的诊断和治疗。因此,通过计算机实现智能影像质控对放射科技师的拍片工作会有较大的辅助作用。本文拟就深度学习领域中的图像分割模型、图像分类模型结合传统图像处理算法应用于医学影像质量评价的研究方法及应用情况予以阐述。我们发现使用深度学习算法对医学影像大数据进行有效训练,提取出来的特征相比于单纯使用传统图像处理算法更加准确、高效,诠释了深度学习在医疗领域的广阔应用前景。本文开发出了一套辅助拍片智能质控系统,并成功应用到了华西医院和其他市、县级医院的放射科,有效验证了该质控系统的可行性与稳定性。     

空气压缩设备故障引起的螺旋断层放射治疗患者计划验证通过率偏低的案例分析

目的:分析影响螺旋断层放射治疗患者计划验证通过率的因素,解决通过率偏低的问题。方法:对以下因素逐一进行验证分析:[①]对ArcCHECK刻度文件重新校准,验证绝对剂量准确性;[②]对ArcCHECK矩阵校准文件重新校正,验证矩阵一致性;[③]利用第三方Matlab平台开源代码进行离线验证,测量计划系统剂量计算准确性;[④]通过每日输出点剂量测量和TomoDose测量,验证加速器输出量稳定性;[⑤]执行AAPM TG-148号报告第V.B.2.c项目和设计临床上常用的2.5 cm和5 cm铅门（Jaw）宽度的适形计划,验证Jaw到位精度;[⑥]设计临床中常用的2.5 cm和5 cm铅门宽度的适形计划和调强计划,验证多叶准直器（MLC）开关时间精度。结果:执行ArcCHECK刻度文件校准和矩阵重新校准之后,计划测试通过率无明显变化;Matlab离线验证脚本获取计划Gamma通过率,结果均在95%以上;晨检记录显示,加速器固定输出量偏差都小于2%,TomoDose结果表明加速器运行稳定;执行AAPM TG-148号报告第V.B.2.c项目检测Jaw到位精度的结果均在误差范围之内,2.5 cm与5 cm不同铅门宽度条件下适形计划,Gamma测试结果以及点剂量测量结果无明显差别;不同计划下点剂量验证误差随靶区形状不规则程度、计划复杂程度和Jaw增加而升高,证明MLC开关时间精度为主要故障因素。结论:排除多种影响因素后最终确定空气压缩设备故障引起的MLC开关时间精度问题才是导致验证通过率较差的主要原因。因此定期维护和保养空气压缩设备十分必要,同时在物理师执行质量保证项目时,应将空气压缩设备内容加入其中。     

A quality assurance program for the on-board imagers

To develop a quality assurance (QA) program for the On-Board Imager (OBI) system and to summarize the results of these QA tests over extended periods from multiple institutions. Both the radiographic and cone-beam computed tomography (CBCT) mode of operation have been evaluated. The QA programs from four institutions have been combined to generate a series of tests for evaluating the performance of the On-Board Imager. The combined QA program consists of three parts: (1) safety and functionality, (2) geometry, and (3) image quality. Safety and functionality tests evaluate the functionality of safety features and the clinical operation of the entire system during the tube warm-up. Geometry QA verifies the geometric accuracy and stability of the OBI/CBCT hardware/software. Image quality QA monitors spatial resolution and contrast sensitivity of the radiographic images. Image quality QA for CBCT includes tests for Hounsfield Unit (HU) linearity, HU uniformity, spatial linearity, and scan slice geometry, in addition. All safety and functionality tests passed on a daily basis. The average accuracy of the OBI isocenter was better than 1.5 mm with a range of variation of less than 1 mm over 8 months. The average accuracy of arm positions in the mechanical geometry QA was better than 1 mm, with a range of variation of less than 1 mm over 8 months. Measurements of other geometry QA tests showed stable results within tolerance throughout the test periods. Radiographic contrast sensitivity ranged between 2.2% and 3.2% and spatial resolution ranged between 1.25 and 1.6 lp/mm. Over four months the CBCT images showed stable spatial linearity, scan slice geometry, contrast resolution (1%; <7 mm disk) and spatial resolution (>6 lp/cm). The HU linearity was within +/-40 HU for all measurements. By combining test methods from multiple institutions, we have developed a comprehensive, yet practical, set of QA tests for the OBI system. Use of the tests over extended periods show that the OBI system has reliable mechanical accuracy and stable image quality. Nevertheless, the tests have been useful in detecting performance deficits in the OBI system that needed recalibration. It is important that all tests are performed on a regular basis.     

呼吸时相和模体位置对4D-CT图像均匀性的影响

目的:通过分析Catphan模体CTP486模块的四维CT（4D-CT）图像,研究呼吸时相和模体位置对4D-CT图像均匀性的影响。方法:使用西门子Sensation Open CT模拟机和瓦里安RPM系统,获取Catphan 504模体CTP486模块的4D-CT图像。对3种模体位置情形进行研究。情形A:模体悬空放置;情形T:模体下有一个碳纤维CT平板床;情形B+T:模体下有一个碳纤维固定底板和一个CT平板床。每种情形重复3次测量。使用DoseLab放疗验证软件分析4D-CT图像,得到图像均匀性值（U）。对每一套4D-CT重建10个呼吸时相（0%, 10%,[ ?], 90%）的CT图像序列。对U值进行统计分析和比较研究。结果:4D-CT所有CT序列的U值均小于5 HU。每种情形得到30个U值数据,情形A、情形T和情形B+T 3种情形U值分别为（1.44±0.79）、（1.91±0.91）和（1.99±0.77） HU。统计U值对应的4个边缘感兴趣区（ROI）出现的次数和比例,情形T中ROI 22（9点钟方向）出现13次（占比43.33%）,情形B+T中ROI 23（12点钟方向）出现13次（占比43.33%）。结论:本研究中4D-CT的图像均匀性满足使用要求,不同呼吸时相4D-CT的图像均匀性值不同,模体摆放位置对4D-CT的图像均匀性有一定影响,碳纤维CT平板床和固定底板的存在使4D-CT 的图像均匀性变差。     

On-Board Imager-based MammoSite treatment verification

Contemporary radiation oncology departments are often lacking a conventional simulator due to common use of virtual simulation and recent implementation of image guided radiation therapy. A protocol based on MammoSite method was developed using CT based planning, a Source Position Simulator (SPS) with a Simulator Wire and a linear accelerator based On-Board Imager (OBI) for daily verification. After MammoSite balloon implantation, the patient undergoes a CT study. The images are evaluated for tissue conformance, balloon symmetry, and balloon surface to skin distance according to the departmental procedure. Prior to the CT study the SPS is attached to the transfer tube that in turn is attached to the balloon catheter. The length from the indexer to the first dwell position is measured using the simulator wire with X-ray markers. After the CT study is performed, the data set is sent to the Varian Eclipse treatment planning system (TPS) and to the Nucletron PLATO brachytherapy planning system. The reference digitally reconstructed radiographs (DRRs) of anterior and lateral setup fields are created using Eclipse TPS and are immediately available on the OBI console via the Varian Vision integrated system. The source dwell position coinciding with the balloon center is identified in the CT dataset, followed by the offset calculation, catheter reconstruction, dose points placement and dwell time calculation. OBI fluoroscopy images are acquired and marked as initial. Prior to each treatment fraction balloon diameter and symmetry are evaluated using OBI fluoroscopy and tools available on the OBI console. Acquired images are compared with reference DRRs and/or initial OBI images. The whole process from initial evaluation to daily verification is filmless and does not undermine the precision of the procedure. This verification time does not exceed 10 min. The balloon diameter correlates well (within 1 mm) between initial CT and OBI verification images. The balloon symmetry is defined with 1 mm accuracy using existing OBI console tools. It is feasible to use OBI based simulation for the MammoSite balloon placement evaluation, balloon integrity daily verification, and treatment dwell position coincidence with balloon center. This verification is a rapid process and is an alternative to the conventional simulator based technique. The simulator wire with X-ray markers for the SPS is the recommended tool for the CT based MammoSite procedure.     

Dosimetric validation of the Acuros XB Advanced Dose Calculation algorithm: fundamental characterization in water

A new algorithm, Acuros? XB Advanced Dose Calculation, has been introduced by Varian Medical Systems in the Eclipse planning system for photon dose calculation in external radiotherapy. Acuros XB is based on the solution of the linear Boltzmann transport equation (LBTE). The LBTE describes the macroscopic behaviour of radiation particles as they travel through and interact with matter. The implementation of Acuros XB in Eclipse has not been assessed; therefore, it is necessary to perform these pre-clinical validation tests to determine its accuracy. This paper summarizes the results of comparisons of Acuros XB calculations against measurements and calculations performed with a previously validated dose calculation algorithm, the Anisotropic Analytical Algorithm (AAA). The tasks addressed in this paper are limited to the fundamental characterization of Acuros XB in water for simple geometries. Validation was carried out for four different beams: 6 and 15 MV beams from a Varian Clinac 2100 iX, and 6 and 10 MV 'flattening filter free' (FFF) beams from a TrueBeam linear accelerator. The TrueBeam FFF are new beams recently introduced in clinical practice on general purpose linear accelerators and have not been previously reported on. Results indicate that Acuros XB accurately reproduces measured and calculated (with AAA) data and only small deviations were observed for all the investigated quantities. In general, the overall degree of accuracy for Acuros XB in simple geometries can be stated to be within 1% for open beams and within 2% for mechanical wedges. The basic validation of the Acuros XB algorithm was therefore considered satisfactory for both conventional photon beams as well as for FFF beams of new generation linacs such as the Varian TrueBeam.     

医学图像三维重建系统的数据结构表达及表面模型的构建

医学图像三维重建在诊断、放射治疗规划及医学研究中均有着重要应用，本文论述了医学图像三维重建系统程序流程，设计了自动及手工轮廓勾画两种分割方法，并提出了建立了合理的系统数据结构。该数据结构能较好地描述系统数据的层次关系和表达重建的几何模型。对由自动分割和手工勾画出的组织，用MT算法构建其三维表面几何模型 。实现了网格简化的边收缩算法，并对由MT算法生成的表面模型进行了网格简化处理。模型网格经简化90％，依然能较好地保持模型的特征，大大加快了绘制速度。     

An Automatic Image Processing Workflow for Daily Magnetic Resonance Imaging Quality Assurance

The performance of magnetic resonance imaging (MRI) equipment is typically monitored with a quality assurance (QA) program. The QA program includes various tests performed at regular intervals. Users may execute specific tests, e.g., daily, weekly, or monthly. The exact interval of these measurements varies according to the department policies, machine setup and usage, manufacturer’s recommendations, and available resources. In our experience, a single image acquired before the first patient of the day offers a low effort and effective system check. When this daily QA check is repeated with identical imaging parameters and phantom setup, the data can be used to derive various time series of the scanner performance. However, daily QA with manual processing can quickly become laborious in a multi-scanner environment. Fully automated image analysis and results output can positively impact the QA process by decreasing reaction time, improving repeatability, and by offering novel performance evaluation methods. In this study, we have developed a daily MRI QA workflow that can measure multiple scanner performance parameters with minimal manual labor required. The daily QA system is built around a phantom image taken by the radiographers at the beginning of day. The image is acquired with a consistent phantom setup and standardized imaging parameters. Recorded parameters are processed into graphs available to everyone involved in the MRI QA process via a web-based interface. The presented automatic MRI QA system provides an efficient tool for following the short- and long-term stability of MRI scanners.     

Automatic re-contouring in 4D radiotherapy

Delineating regions of interest (ROIs) on each phase of four-dimensional (4D) computed tomography (CT) images is an essential step for 4D radiotherapy. The requirement of manual phase-by-phase contouring prohibits the routine use of 4D radiotherapy. This paper develops an automatic re-contouring algorithm that combines techniques of deformable registration and surface construction. ROIs are manually contoured slice-by-slice in the reference phase image. A reference surface is constructed based on these reference contours using a triangulated surface construction technique. The deformable registration technique provides the voxel-to-voxel mapping between the reference phase and the test phase. The vertices of the reference surface are displaced in accordance with the deformation map, resulting in a deformed surface. The new contours are reconstructed by cutting the deformed surface slice-by-slice along the transversal, sagittal or coronal direction. Since both the inputs and outputs of our automatic re-contouring algorithm are contours, it is relatively easy to cope with any treatment planning system. We tested our automatic re-contouring algorithm using a deformable phantom and 4D CT images of six lung cancer patients. The proposed algorithm is validated by visual inspections and quantitative comparisons of the automatic re-contours with both the gold standard segmentations and the manual contours. Based on the automatic delineated ROIs, changes of tumour and sensitive structures during respiration are quantitatively analysed. This algorithm could also be used to re-contour daily images for treatment evaluation and adaptive radiotherapy.     

基于特征提取量化分析的体外活细胞追踪算法研究

目的:提高显微镜下序列图像细胞追踪的效率及准确度。方法:提出双阈值形态学与拓扑约束图论法相结合的 自动细胞追踪算法,用来分析体外活细胞定向迁移轨迹及参数,并从细胞数目及细胞特征两方面分析追踪算法的准确 性。在特征分析方面,从运动速度、运动距离、趋化速度、趋化指数和方向持续性5个指标与手动采样数据进行对比。结 果:该算法可以分别识别在毛细管针部灰度较高区域的细胞及其他区域灰度较低的细胞,细胞数目准确度平均达到 91.8%,分析得到的5个特征指标与手动采样分析结果基本一致,误差不超过5%。结论:双阈值形态学与拓扑约束图论法 相结合的自动细胞追踪算法可以有效提高细胞追踪的准确度。     

一种关于RPM放疗呼吸门控系统束流时间延迟性能的参考质控方法

目的：提出一种关于RPM放疗呼吸门控系统束流时间延迟性能的参考质控方法,给出参考条件下VB与EDGE加速器的时间延迟测量结果。方法：首先设计梯形质控呼吸曲线并加载运动模体,利用定位CT扫描由运动模体驱动的W-L模体,然后设计用于束流出束和截止延迟测量的两类质控计划,最后在加速器上执行质控计划,使用EPID采集模体的静态参考图像与运动测量图像,通过分析致密金属球在运动图像和参考图像中的位置差,反推束流的时间延迟,分别在VB与EDGE两台加速器上做方法验证。结果：VB与EDGE加速器的束流出束时间延迟均小于100 ms,截止时间延迟基本一致,VB加速器约为14 ms,EDGE约为22 ms。10FFF各剂量率的出束时间延迟基本一致,而6FFF、6 MV和10 MV则随剂量率的变化略有增加。4个能量各剂量率的束流截止时间延迟均较小且相对一致,部分能量有随剂量率变大而延迟缩小的趋势。结论：本研究提出的RPM束流时间延迟参考质控方法和条件具有较高的测量可信度和较强的临床实操性,测量结果表明RPM呼吸门控系统响应灵敏,研究结果为呼吸门控系统的时间延迟质控提供了重要的方法学指导与数据参考。