Matovic Milovan, Jankovic Milica, Barjaktarovic Marko, Jeremic Marija
Clinical Center Kragujevac, Center of Nuclear Medicine, Serbia.
Hell J Nucl Med. 2017 Sep-Dec;20 Suppl:159.
After radioiodine therapy of differentiated thyroid cancer (DTC) patients, whole body scintigraphy (WBS) is standard procedure before releasing the patient from the hospital. A common problem is the precise localization of regions where the iod-avide tissue is located. Sometimes is practically impossible to perform precise topographic localization of such regions.
In order to face this problem, we have developed a low-cost Vision-Fusion system for web-camera image acquisition simultaneously with routine scintigraphic whole body acquisition including the algorithm for fusion of images given from both cameras. For image acquisition in the gamma part of the spectra we used e.cam dual head gamma camera (Siemens, Erlangen, Germany) in WBS modality, with matrix size of 256×1024 pixels and bed speed of 6cm/min, equipped with high energy collimator. For optical image acquisition in visible part of spectra we have used web-camera model C905 (Logitech, USA) with Carl Zeiss® optics, native resolution 1600×1200 pixels, 34 field of view, 30g weight, with autofocus option turned "off" and auto white balance turned "on". Web camera is connected to upper head of gamma camera (GC) by a holder of lightweight aluminum rod and a plexiglas adapter. Our own Vision-Fusion software for image acquisition and coregistration was developed using NI LabVIEW programming environment 2015 (National Instruments, Texas, USA) and two additional LabVIEW modules: NI Vision Acquisition Software (VAS) and NI Vision Development Module (VDM). Vision acquisition software enables communication and control between laptop computer and web-camera. Vision development module is image processing library used for image preprocessing and fusion. Software starts the web-camera image acquisition before starting image acquisition on GC and stops it when GC completes the acquisition. Web-camera is in continuous acquisition mode with frame rate f depending on speed of patient bed movement v (f=v/∆, where ∆ is a displacement step that can be changed in Settings option of Vision-Fusion software; by default, ∆ is set to 1cm corresponding to ∆=15 pixels). All images captured while patient's bed is moving are processed. Movement of patient's bed is checked using cross-correlation of two successive images. After each image capturing, algorithm extracts the central region of interest (ROI) of the image, with the same width as captured image (1600 pixels) and the height that is equal to the ∆ displacement in pixels. All extracted central ROI are placed next to each other in the overall whole-body image. Stacking of narrow central ROI introduces negligible distortion in the overall whole-body image. The first step for fusion of the scintigram and the optical image was determination of spatial transformation between them. We have made an experiment with two markers (point radioactivity sources of Tc pertechnetate 1MBq) visible in both images (WBS and optical) to find transformation of coordinates between images. The distance between point markers is used for spatial coregistration of the gamma and optical images. At the end of coregistration process, gamma image is rescaled in spatial domain and added to the optical image (green or red channel, amplification changeable from user interface).
We tested our system for 10 patients with DTC who received radioiodine therapy (8 women and two men, with average age of 50.10±12.26 years). Five patients received 5.55Gbq, three 3.70GBq and two 1.85GBq. Whole-body scintigraphy and optical image acquisition were performed 72 hours after application of radioiodine therapy.
Based on our first results during clinical testing of our system, we can conclude that our system can improve diagnostic possibility of whole body scintigraphy to detect thyroid remnant tissue in patients with DTC after radioiodine therapy.
在分化型甲状腺癌(DTC)患者接受放射性碘治疗后,全身闪烁扫描(WBS)是患者出院前的标准程序。一个常见问题是碘亲和组织所在区域的精确定位。有时几乎不可能对这些区域进行精确的地形定位。
为了解决这个问题,我们开发了一种低成本的视觉融合系统,用于网络摄像头图像采集,同时进行常规闪烁扫描全身采集,包括用于融合两台相机图像的算法。对于光谱伽马部分的图像采集,我们在WBS模式下使用e.cam双头伽马相机(西门子,德国埃尔朗根),矩阵大小为256×1024像素,床速为6cm/分钟,配备高能准直器。对于光谱可见光部分的光学图像采集,我们使用型号为C905的网络摄像头(罗技,美国),带有卡尔蔡司光学元件,原生分辨率为1600×1200像素,视野为34,重量为30g,自动对焦选项关闭,自动白平衡打开。网络摄像头通过轻质铝杆支架和有机玻璃适配器连接到伽马相机(GC)的上头。我们使用NI LabVIEW编程环境2015(美国国家仪器公司,德克萨斯州)以及另外两个LabVIEW模块:NI视觉采集软件(VAS)和NI视觉开发模块(VDM),开发了我们自己的用于图像采集和配准的视觉融合软件。视觉采集软件实现笔记本电脑与网络摄像头之间的通信和控制。视觉开发模块是用于图像预处理和融合的图像处理库。软件在GC开始图像采集之前启动网络摄像头图像采集,并在GC完成采集时停止。网络摄像头处于连续采集模式,帧率f取决于患者床移动速度v(f = v / ∆,其中∆是可以在视觉融合软件的设置选项中更改的位移步长;默认情况下,∆设置为1cm,对应于∆ = 15像素)。对患者床移动过程中捕获的所有图像进行处理。使用连续两幅图像的互相关来检查患者床的移动。每次图像捕获后,算法提取图像的中央感兴趣区域(ROI),其宽度与捕获图像相同(1600像素),高度等于以像素为单位的∆位移。所有提取的中央ROI在整体全身图像中彼此相邻放置。窄中央ROI的堆叠在整体全身图像中引入的失真可忽略不计。闪烁图和光学图像融合的第一步是确定它们之间的空间变换。我们用两个在两幅图像(WBS和光学)中都可见的标记(高锝酸盐1MBq的点状放射性源)进行了实验,以找到图像之间的坐标变换。点状标记之间的距离用于伽马图像和光学图像的空间配准。在配准过程结束时,伽马图像在空间域中重新缩放并添加到光学图像(绿色通道或红色通道,放大倍数可从用户界面更改)。
我们对10例接受放射性碘治疗的DTC患者测试了我们的系统(8名女性和2名男性,平均年龄50.10±12.26岁)。5例患者接受了5.55GBq,3例接受了3.70GBq,2例接受了1.85GBq。在应用放射性碘治疗72小时后进行全身闪烁扫描和光学图像采集。
基于我们系统临床测试的初步结果,我们可以得出结论,我们的系统可以提高全身闪烁扫描检测放射性碘治疗后DTC患者甲状腺残留组织的诊断可能性。