【作者】 谌家喜；

【导师】 周荷琴；

【作者基本信息】 中国科学技术大学 ， 模式识别与智能系统， 2008， 博士

【摘要】 随着人们日常生活水平的提高,越来越多的人开始注重医疗保健,国家也投入了大量的人力物力来推进医疗水平的提高。医学影像诊断是一种常规的医疗检查方式,在疾病诊治及企事业单位和个人的体检中得到广泛应用,其中,又尤以X射线成像的检查居多。在材料科学和计算机技术发展的带动下,X射线成像已经从原来的胶片成像走进了直接数字化x射线成像(Direct DigitalRadiography,DR)时代。在更快更好成像的同时,数字化X射线成像也带来了一些问题:曝光系统的自动化控制,X射线到电子信号的转化,成像系统的校正,以及X射线数字图像的处理。对DR系统成像方法进行研究,不仅能改善数字化X射线图像质量,提高医院影像科的疾病诊治水平,还可为其它医学成像模式的研究提供借鉴,更可为进一步的计算机辅助诊断打下基础。由此可见,DR系统成像方法研究具有重要的理论意义和应用价值。本论文从DR系统的图像形成过程出发,研究了相关环节中的关键问题,即高压发生器控制、图像采集和校正、图像后处理,目标是实现DR系统的优化控制,并获取更好的X射线数字图像。高压发生器是X射线球管的电力供给装置,在成像过程中需要和计算机系统有良好的配合,以控制曝光流程并配合曝光参数交换;X射线透射过人体后到达接收装置,根据成像原理和材料特性,图像采集和图像校正主要解决从X射线到计算机数字存储中的能量转换问题;图像后处理是在X射线数字图像的基础上,运用多种图像处理算法,进一步提高图像质量,以满足实际应用的需求。本文对DR系统成像进行了深入的研究,取得了以下成果:(1)通过对原有DR系统控制台接口电路的改造,用PC机串口和相应的软件实现了计算机与高压发生器之间的通信,操作者只需坐在PC机前就可以方便地控制高压发生器,不需要在多个控制台之间进行较为繁复的操作,既加快了摄片的流程,还节省了DR系统的硬件成本,并且为实现DR系统的参数自动化控制打开了新思路。(2)在研究了DR系统图像采集中的信号控制问题后,设计了平板探测器状态控制和数据处理流程;提出了原始图像的隔行扫描及其区域交换算法,解决了如何将原始图像调整为正常的X射线图像的问题,并在此基础上设计了DR图像的采集、显示和存储程序。分别对水模和人体部位进行了X射线成像测试,实验结果表明了该方法的准确性和可靠性,解决了DR图像采集过程中的信号控制和图像处理的问题。获取的X射线数字图像真实有效,并已经被应用于医院的DR系统。(3)结合DR系统成像原理及图像采集过程,从X射线到计算机数字图像过程中的能量转换角度,详细研究了由X射线光场不均匀性和平板探测器的像元响应不一致性带来的图像噪声,并针对各种影响因素,采取了相应的校正方法。最后,针对现场系统的X射线场不可测性,提出了一种新的整体校正理论,设计并实现了一种行之有效的现场操作算法。(4)研究了多幅DR图像的拼接问题,提出了基于特征点的图像配准算法,并使用加权平均法对图像进行了融合,解决了DR图像拼接中亮度相差较大、相似特征点较多和位移较大以及处理时间较长等问题,并在此基础上设计了DR图像的拼接算法程序。分别对物体和动物鱼以及人体进行了X射线摄影,做了图像拼接实验,取得了较好的拼接效果。(5)研究了X射线图像中的曝光区域特性,提出了X射线图像曝光区域自动检测出算法,并进一步对图像进行自动裁剪,减少了图像的存储空间,方便于医生的操作和诊断。在此基础上,提出了一种可伸缩性图像增强算法,使得图像整体和细节都有较好的改善,进一步提高图像质量;最后,采用优化的窗宽窗位算法来显示图像。更多还原

【Abstract】 With the rapid improvement of the daily life, more and more people begin to pay attention to the health care. The state has also invested a lot of manpower and material resources to promote the health care standards. Medical image diagnosis is a kind of routine medical examinations. It has been widely used in the diagnosis and treatment of disease and body check-up of the enterprises and institutions or the individuals. Especially, the X-ray imaging examination is frequently used. With the development of the materials science and the computer technology, the X-ray imaging technology has entered the direct digital radiography (DR) era from the original film. It takes faster imaging speed and better image quality, and also the digital X-ray imaging has brought out some new problems. For example, how to control the exposure system automatically? What’s the conversion from the X-ray to the electronic signals? The imaging system must be calibrated. And the digital X-ray images need to be post processed. By studying the imaging method of the DR system, the quality of the digital X-ray image can be improved, and the diagnosis and treatment of diseases in the hospital will be boosted. It is also a reference for the study of other medical imaging models. More over, it is the foundation for the computer-aided diagnosis in the future. Evidently, it is very important to study the imaging method for the DR system.Based on the imaging process of the DR system, several key problems of the relevant part have been studied in this thesis. Such as the control of the high-voltage generator, image acquisition and calibration, image post-processing. Our goal is to achieve optimal control of the DR system and get better quality X - ray images. The high-voltage generator is the power-supply device for the X-ray, which needs to have good cooperation with the computer system so as to control the exposure process and exchange exposure parameters for the X-ray imaging. The X-ray arrives at the receiving device after transmission involving the human body. Based on the imaging principle and material properties, the image acquisition and image correction mainly solve the energy conversion problems from the X-ray to the computer. After having achieved the X-ray images, we utilize a variety of image processing algorithms to further improve the image quality, so as to satisfy the applications. We have carried out in-depth research on the DR imaging system, and several productions have been achieved as following:(1) A proposal is presented to improve the original complicated control of the exposure parameters on the DR system. We redesigned the interface circuitry of the high-voltage generator by using RS-232 serial COM port. The data and signals are transmitted between personal computer and high-voltage generator with serial mode. A program package is designed to realize the control of the exposure parameters for high-volt generator of DR on the PC machine. Experiment results shown that the proposed console operates steadily and it is possible to provide a more convenient operation console for the direct radiography system.(2) The signal control strategy in a Matrox Meteor II/Camera Link card-based DR image acquisition system has been studied. We designed the processing flow for status control and data process of the flat panel. A line interval scan and area exchange algorithm is presented to process the original image data acquired from the detector array. The problem on the adjustment of the abnormal original image has been solved successfully, and a program is developed for DR image acquisition, display and storage. Experimental results on water models and human bodies show that the methods described here were correct and credible, and the digital x-ray images were valid. It has been applied in the clinical practices of the DR system.(3) Based on the principle of DR imaging and the image acquisition process, we take our point of view on the energy conversion perspective from the X-ray to the computer. A detailed study has been carried out on the image noise produced by the inhomogeneity of the X-ray field and the flat panel detector pixel response. According to the various factors, corresponding correction methods have been introduced. Finally, for the uncertainty of the X-ray system at use, we extract a new theory for the overall correction, which has been designed and implemented as an effective field operation algorithm.(4) The problem of several DR images stitching has been studied. An image registration algorithm based-on feature points is presented. And a weighted average algorithm is used to solve the image fusion process. There are many embarrassments in the stitching process, such as light difference, too many similar feature points, big displacement, and long time for process. In our approach, these problems are well-handled. The programme has been complied to carry out the algorithms. We have taken the X-ray images for objects, fish and the human body. Experimental results show that the methods described here were correct and credible, and the composed digital x-ray image was valid. It can be applied in the design of a direct digital radiography system.(5) The exposure region characteristics of the X-ray image have been studied. An automatic algorithm has been introduced for detecting and cutting the exposure region of the image, which reduced the image storage space and facilitate the doctors operation and diagnosis. On this basis, a scalable image enhancement algorithm was imported, which makes better improvement in the overall and details in the image. The image has been further enhanced. Finally, we optimized the so-called window width and window level algorithm to display the images.更多还原