医学图像有损压缩技术研究进展

医学成像已经成为医学诊断与处理的重要依据。随着医学成像设备分辨率的不断提高,所获取的医学图像数据量也在持续增长,这给医学图像的存储与实时传输带来了巨大压力,并严重制约了其后续应用。有损压缩方式能够在满足一定图像质量的条件下,实现医学图像较大程度的压缩,目前已成为国内外研究的热点。本文对医学图像有损压缩技术研究进展进行总结,并介绍有损压缩条件下的质量评价方法,最后对医学图像有损压缩技术的发展趋势进行展望。     

医学图像查看系统Radworks的应用体会

目的：探讨医学图像查看系统Radworks的应用。材料与方法：通过Applicare公司开发的医学成像和PACS软件Radworks5．0系统获取、归档和传输放射科影像资料。结果：为了兼顾各种新旧不同的检查设备,部分图像的获取是通过有DICOM接口的新设备直接传输获取,而另一部分图像的获取是通过专用胶片扫描仪扫描X线照片获得,用光盘做数据的离线备份。结论：对于数据量不大、规模小的微型PACS,Radworks5．0尚可满足图像的获取、归档与传输等功能。     

医学图像的零树小波编码

背景:医学数字图像必须是高质量的、高分辨率,所以数据量很大,如此巨大的数据量不利于图像存档与传输系统的运行和数字化医院、远程医疗的实现.因此,图像压缩成为图像存档与传输系统要解决的重要问题.目的:分析零树小波变编码算法原理并编程实现对医学数字图像的压缩,使之能够满足医学图像的传输和诊断要求.方法:应用嵌入式零树小波编码算法,探讨小波基和小波变换层数的选择,编程实现对医学数字图像的压缩.结果与结论:选择双正交小波基对医学图像进行4层小波变换实现压缩,获得了较高的峰值信噪比,取得了较好的压缩效果.     

压缩感知技术及其在MRI上的应用

压缩感知是基于应用数学的一种创新的信号获取及处理理论,其原理是通过对所采集的信号进行适当域变换得到可压缩信号,直接采集压缩后的信号并利用重构算法实现快速优质信号重建。运用该技术成像不仅具有出色的时间分辨率优势,同时具有满意的空间分辨率,因此近年来其在医学成像领域的应用逐渐成为研究热点。作者在阐述压缩感知理论基本原理的基础上,进一步对其在MRI上的研究现状和发展前景进行综述。     

数字医学影像设备综述

随着医学影像技术近几十年的不断发展,DR、CT、磁共振以及分子成像等技术与设备研发取得十足的发展。结合pacs系统,数字医学影像设备在医学临床中的应用进入数字化、智能化时代。快速、清晰的图像获取,辅助医生更好地判断患者的疾病类型,并准确找到病因,从而进行精准治疗。提升诊断水平的同时,推动医学的发展,为人类的健康做出极大贡献。     

相控阵超声诊断仪中对放的原理及设计

高质量的医学图像对临床诊断是十分有利的,影响医学图像质量的因素很多。对B型超声断层成像设备而言,若使诊断仪获得的医学信息不丢失地显示出来,丰富图像的信息量,就要对回波信号进行有效的压缩处理。本文介绍了实现这种压缩处理功能的电路—对数放大器的原理及设计。     

有损压缩效应在胸部X 线照片的微细异常检测

远程放射会诊目前已成为公认的放射实践的组成部分。医学影像的庞大容量的解决方法是主要因素 ,影像容量问题只能是压缩图像 ,减少影像尺寸 ,无损压缩法在使用了图像内的信息量的同时减少了图像资料的容量从而可以完整的重建图像。这种无损技术明显的优点是完整的使用图像重建。但医学影像只能压缩到 2 :1～ 3:1。不可逆的或有损技术能按任何设计的比例压缩 ,但压缩比例越高 ,复制的图像与原始图像失真越大。这就促使我们去评价医学图像有损压缩的效果。由有损压缩产生的图像降级可以由数字计算进行测量。但是 ,以往所作的关于压缩对图像影…     

基于Visual C++和Matlab的磁共振影像增强后处理系统

目的:随着医学影像学的飞速发展,手术导航技术的应用及脑功能等图像分析研究的不断深入,基于医学数字成像和通信标准的医学影像分析与处理也随之成为医学图像处理领域中的热点。为便于科研人员研究相应的磁共振图像局部增强等后处理算法及进行图像分析,提出一种基于VisualC 和Matlab的磁共振影像增强后处理研发平台。方法:对北京协和医院放射科2005-01/10获取的部分磁共振医学数字成像和通信图像利用灰度扩展进行全局增强,利用基于数学形态学方法进行局部增强算法的研究。在程序实现上使用Matlab引擎实现VC 和Matlab混合编程处理医学数字成像和通信图像。结果:为磁共振图像进行增强局部对比度算法研究提供了一个研发平台,实现了位图图像与医学数字成像和通信图像的数据转换接口功能。结论:处理后的图像具有更好的应用价值,为图像局部对比度增强算法的研究提供一个有效的平台。在算法研制阶段采用VC和Matlab混合编程的方法可以提高算法研究效率。     

超声与CT/MRI图像融合技术的临床应用

将几种医学成像设备获取的图像，经过必要的变换处理，达到空间坐标上的匹配，叠加后获取互补信息，可弥补单一模式成像的某些不足。图像融合不仅使临床诊断和治疗更加准确和完善，而且在肿瘤物理治疗、立体定向放射外科、图像引导的手术导航系统、医学图像的归档与通信系统及远程医学中有着重要意义。     

数字化医学成像设备共性技术分析

医学成像设备通过获取人体的物理信息进行处理和成像,给疾病的诊治提供重要参考。本文通过分析数字化医学成像设备的一般处理过程,对其中的共性技术及其进展做一介绍。     

Iterative reconstruction for coronary CT angiography: finding its way

Image reconstruction algorithms play a critical role in defining the quality and integrity of medical imaging using computed tomography. Since the advent of CT, image reconstruction has largely been performed by filtered back projection (FBP). This reconstruction technique has served CT well particularly at a time when there were significant limitations in computer processing capabilities. Iterative image reconstruction algorithms were, in fact available and were used to generate images with the very first commercial clinical computed tomographic (CT) scanner. This technique did not see significant adoption in clinical CT use owing to the ease of implementation and the faster image reconstruction of filtered back projection. Over the past decade, the need for finer resolution, greater volume coverage, faster scan times and the desire to lower radiation dose at the same time have pushed the performance of FBP reconstruction to its limits. Recently, there has been a re-introduction of iterative reconstruction for CT imaging with recently published studies in other organ systems showing that iterative reconstructions can produce higher-resolution images with greater robustness for the reduction of various imaging artifacts. There has been subsequent early adoption and experience with iterative reconstruction in coronary CT angiography (CCTA). We herein review the various iterative reconstruction platforms released for use for CCTA and the initial experiences implementing and integrating these reconstruction algorithms in clinical practice.     

Magnetic induction tomography: experimental realization

Magnetic induction tomography (MIT) is a new non-contacting technique for visualization of the electrical impedance distribution inside inhomogeneous media. A measuring system for MIT has been developed. An oscillating magnetic field is applied in the system as a sounding agent. The system is designed mainly for biomedical applications. Experiments demonstrate that with proper selection of measurement conditions it is possible to use the phase shifts between inductor and detector signals for image reconstruction by filtered backprojection along magnetic lines. Measurements with saline filled phantoms having various spatial distributions of conductivity were carried out and images were reconstructed. The experiments have demonstrated the applicability of MIT for medical imaging and diagnostics.     

Real time tissue elasticity imaging using the combined autocorrelation method

The elastic properties of tissues are expected to provide novel information for use in diagnosing pathologic changes in tissues and discriminating between malignant and benign tumors. Because it is hard to directly estimate the elastic modulus distribution from echo signals, methods for imaging the distribution of tissue strain under static compression are being widely investigated. Imaging the distribution of strain has proven to be useful for detecting disease tissues on the basis of their differences in elastic properties, although it is more qualitative than elastic modulus distribution. Many approaches to obtaining strain images from echo signals have been proposed. Most of these approaches use the spatial correlation technique, a method of detecting tissue displacement that provides maximum correlation between the echo signal obtained before and the one obtained after compression. Those methods are not suited for real-time processing, however, because of the amount of computation time they require. An alternative approach is a phase-tracking method, which is analogous to Doppler blood flowmetry. Although it can realize the rapid detection of displacement, the aliasing effect prevents its application to the large displacements that are necessary to improve the S/N ratio of the strain image. We therefore developed a more useful technique for imaging tissue elasticity. This approach, which we call the combined autocorrelation (CA) method, has the advantages of producing strain images of high quality with real-time processing and being applicable to large displacements. Numeric simulation and phantom experimentation have demonstrated that this method's capability to reconstruct images of tissue strain distribution under practical conditions is superior to that of the conventional spatial correlation method. In simulation and phantom experimentation, moreover, the image of elastic modulus distribution was also obtained by estimating stress distribution using a three-dimensional tissue model. When the proposed CA method was used to measure breast tumor specimens, the obtained strain images clearly revealed harder tumor lesions that were only vaguely resolved in B-mode images. Moreover, the results indicated the possibility of extracting the pathological characteristics of a tumor, making it useful for determining tumor type. These advantages justify the clinical use of the CA method.     

基于整数小波变换和零树编码算法的医学图像压缩研究

现代医疗成象设备产生了大量高价值的医学图像,如何对信息的进行有效的存储、查询以及网络传输是一个亟待解决的问题.本文利用整数小波变换和零树编码算法对医学图像进行了压缩研究,试验表明,同传统JPEG标准相比,相同压缩比下本算法的峰值信噪比有明显的提高,同时本算法具有逐渐显现的特性,能够满足医学图像存储、查询以及网络传输的需求.     

Tissue Doppler, a fundamental tool for parametric imaging

Tissue Doppler has been used for clinical applications since 1989. It has been developed from a pulsed Doppler acquisition tool towards a method where extraction of velocities can be performed from colour-coded images. This has introduced a further development into different forms of parametric images describing different myocardial functions as colour-coded information, like deformation imaging, motion imaging and phase imaging. The technical requirements have been established with temporal requirements of frame rates in acquisition exceeding 100 frames s(-1). The most powerful application of the tissue Doppler technique today is perhaps to quantify the myocardial functional reserve, during stress echocardiography, making the method applicable to diagnose the presence of coronary disease with an accuracy exceeding that of nuclear and other non-invasive techniques. The method has also great potential for future developments with introduction of more regional measuring variables.     

Clinical Outlines

One of the most exciting developments in medicine within the past decade has been the application of nuclear magnetic resonance (NMR) imaging to clinical diagnosis. Although medical use of NMR imaging is still in its infancy, this revolutionary procedure already competes favorably with computed tomography and ultrasound. It not only produces high-resolution three-dimensional images of the body safely and noninvasively but also has the potential to allow evaluation of tissue pathology and metabolic function. Dr Vannier, who is a pioneer in application of this space-age technology to medicine, explains the important advantages of NMR imaging, describes its spectacular achievements to date, and counsels physicians on its clinical applications.     

人工智能在超声医学领域中的应用

人工智能(artificial intelligence,AI)近几年再度成为各领域关注的焦点,其中深度学习的提出带来了一系列革命性变化,而随着计算机视觉向深度学习过渡以及硬件和大数据的进步,AI在图像识别领域展现出更广阔的发展前景。深度学习模型使得相关图像算法甚至达到了比人眼更高的识别准确率, 这为医学影像的发展提供了巨大契机。超声医学作为影像领域的重要分支,利用AI相关算法进行声像图分析的研究不断涌现,不仅为临床科研提供了新思路,亦有助于提高超声诊断的准确性。     

多模态深度学习及其在眼科人工智能的应用展望

深度学习的强学习能力和高易用性使其成为当前主流机器学习算法和医学人工智能的核心技术。鉴于医学影像在健康筛查、疾病诊断、精准治疗、预后评估等诸多任务中的关键作用,用于医学影像结构分析与语义理解的深度学习正成为重要的交叉学科研究方向。在临床场景中,医生为了实现更精准的诊断,往往需要同时参考不同类型、不同模态的影像样本进行综合分析和判断。本文介绍面向此类场景的多模态深度学习的基本概念和工作原理,结合具体案例分析多模态深度学习在眼科领域的研究进展、应用情况及技术挑战,并对该技术的应用前景作出展望。