Rhythmic is a basic element of human life.During the fast repetitive movements cortical potential represents different characters from those of separate movements performed slowly.In this article,we record the electrocorticographic(ECG) activity from the brain surface of two epileptic patients during fast repetitive movement in two kinds of rates,analysis movement-related power and movement-related coherence of ECG signals and uncover that the power on special frequency band goes up while the coherence goes down.The two indexes appear statistically significant difference between fast and slow rate movement.Analysis of the coherence between different functional regions shows that SMA may be related to the movement rate.

Previous research on steady-state visual evoked potential(SSVEP) and inverse EEG shows amplitude and spatial differences of visual evoked response to steady-state visual stimulus of different frequencies.The fMRI experiments were employed to study the frequency dependence of visual evoked response in this paper.The experiments were designed according to SSVEP results,5 volunteers were scanned on MR scanner while they were presented visual stimulus of several frequencies to acquire fMRI data.Results of pre-processing and statistical analysis by SPM show unilateral dominance of response regions which displays the dominance representation in steady-state visual response.Possibility of relationship between the unilateral dominance and stimulus properties is discussed,as well as the perspectives of further research and application.

In epilepsy surgery,electrodes are often implanted between the skull and brain or inside the brain to record the EEG for locating the seizures onset zone.After the implantation,clinicians usually have to find out the positions of these electrodes with respect to brain structure according to MRI scans.However,the existing locating methods either need much user assistance or are low efficient.We developed an efficient automatic method that could locate the implanted electrodes efficiently with little user interaction.This method makes use of the geometrical character as well as the appearance of electrode grids in MRI images first,and then regularize the result using a smooth curved surface model.Finally these electrodes are visualized in a 3D reconstructed brain model to help the surgery planning.This method was applied on real MRI images and the results proved that the method was very effective for locating implanted electrode grids in MR images.

To address the issue of system simplicity and subject applicability, a brain controlled HCI system derived from steady state visual evoked potential (SSVEP) based brain computer interface (BCI) is proposed in this paper. Aiming at an external input device for personal computer, key issues of hardware and software design for better performance and user-friendly interface are introduced systematically. With proper parameter customization for each individual, an average information transfer rate of 46bits/min was achieved in the operation of dialing a phone number. With encouraging online performance and advantages of system simplicity, the proposed HCI using SSVEP-based BCI technology is promising for a substitute of standard computer input device for both health and disabled computer users.

The motivation of the brain-computer interface (BCI) research and its potential applications are introduced in this paper. Some of the problems in BCI-based medical device developments are also discussed.

A linear recognition method was developed to analyze brain DT-MRI(diffusion tensor magnetic resonance imaging) of Alzheimer's disease for computer aided diagnosis.Nine Alzheimer's patients and nine controls received DT-MRI scanning in a 1.5 T MRI system.The appearance diffusion coefficient and the fractional anisotropy were measured for the splenium of the corpus callosum,the genu of the corpus callosum,and the hippocampus.These 2 parameters of the 3 fields were used to statistically train the model.The cross validation method with leave-one-out was used to estimate the recognition rate,with an accuracy of 83.3%. The results show that the recognition method can assist doctors with Alzheimer's diagnosis and that DT-MRI can reflect the abnormalities of Alzheimer's patients brains.

Geometric distortion is one of the most important criteria for evaluating computer radiography (CR) imaging quality. A two-step sort order method was developed to calculate the elliptical geometrical distortion in an X-ray image. The method was used to calculate the geometric distortion of CR mammogram images. The measured data was then compared with standard mammograms with the same imaging conditions. Analysis of 56 CR images indicated that the geometrical distortion was less then the geometrical distortion detectable by the human eye. The results demonstrate that CR images are suitable for diagnosis of breast cancer.

在人体骨骼系统的三维可视化过程中,将二维CT数据重建生成理想的三维骨骼数据是非常重要的一环.作者应用可视化工具Visualization Toolkit(VTK)和VC++实现了二维数据的三维重建、面片削减、连通性检测、三维平滑等一系列功能.与利用OpenGL等工具重建得到的骨骼数据相比,不但准确度和视觉效果有了显著改善,数据量也大大减小.VTK在人体骨骼系统的重建及处理工作中取得的良好效果表明,它是医学图像重建和处理的有力工具之一,具有使用灵活,功能强大的优点,值得推广.

We have constructed a three-layer B-P artificial neural network based on the physician's diagnosis.Twenty templates are used to train the network.In addition,we use this network to make pattern recognition for lung cancer in X-ray radiography.The operating feature of this network is also tested.Through our study,we find the practical and potential advantage of application ANET in medical imaging pattern recognition.

血流测量超声诊断中的重要内容.本文介绍了三维超声血汉测量的意义、工程实现方法及最新的研究进展.

根据微循环图象处理和参数测量的发展的需要, 研制开发了MCIPPMS-微循环图象处理及参数测量系统(VERSION5.0), 其软件系统是在WINDOWS95/WINDOWS98操作系统下运行的WINDOWS中文版本. 该系统主要用于显微摄像的微循环图象分析处理和参数测量, 也可用于其他医学图象或一般通用图象的分析和处理. 该系统由硬件系统和软件系统二部分组成: 其硬件系统由微循环镜、摄像机、录像机、图象采集片、微型计算机、打印机等组成. 其软件系统在WINDOWS95/WINDOWS98操作系统下运行, 全中文界面. 主菜单由文件、编辑、图象分析、图象处理、参数测量、图形图表、数据库、网络、窗口、帮助等项及其相应程序组成. 图象输入可由实时视频采集, 文件读入, 并支持与其他数字化图象设备的接口. 图象输出可存盘、打印或发送. 图象管理和编辑部分, 可完成图象的输入和输出管理、视频图象的实时采集和显示、图象观察的放大、缩小或旋转以及图象尺寸的调整、图象的剪切编辑、缩略图象预览等功能. 图象分析和图象处理部分, 配置有图象分析和图象处理的多种程序、图特征分析、图象增强、多种图象模块滤波、图象频域滤波、图象二值化、伪彩色图象、区域填充和分割、边缘或轮廓的检出、二值图象的腐蚀、膨胀和细化等. 在该部分的设置项中, 包含多种自动或手动的预设置选项, 以适合各种不同对象的要求. 该系统中微循环多种参数的测量, 特别是动态参数测量, 主要基于对窗口动态序列图象的分析和处理的方法而进行的. 测量参数包括: 一维时-空血流速度场、二维速度场、时-空血管直径、面积、长度、通透性测量、血管网测量、细胞提取和测量、动力学特性分析和测量等. 测量结果可通过图形图表进行显示、存盘或打印. 图形图表项包括对一维二维图数据和测量结果的多种形式的图表显示. 该系统的数据库, 可进行图象库的存取记录、测量结果的存取记录, 并可根据用户的情况, 实现对本地或远程数据库的访问和存取. 在帮助菜单下, 提供了该系统下各项操作方法的说明和帮助. 该系统的软件是面向对象的模块化结构, 可根据用户的需求开发增添更多的功能.

对微血管中的血流运动的测量，人们已经进行过较长时间的研究，提出了飞点扫描、双窗法、单窗法、激光多普勒测速等测量方法［1］。属于图像分析类的方法主要有相关法、时空梯度法以及方向滤波法等。本文所关心的是图像分析方法。

为测试溶栓药物效果, 设计实验如下: 不损伤仓鼠硬脑膜情况下开颅, 使用红外激光灼伤仓鼠脑部, 形成微白血栓, 溶入溶栓药物, 利用特制微循环观察设备取得仓鼠脑部微循环图象, 然后利用图象卡采集变成数字化图象, 送入计算机处理. 由于获取的图象质量很差, 而实验要求从图象处理的角度给出一个可测的指标, 直观的反应血栓在溶栓过程中的变化状况. 因而通过仔细观察灰度图象的直方图, 发现: 图象中存在比较明显的双峰现象, 也就是说图象中血液部分灰度和血栓部分灰度分别集中于二个灰度值附近, 在这二个峰值附近存在明显的分界谷底灰度. 基于以上观察, 采用图象切割和模式识别的方法来将这二种灰度区分开来. 首先将包含血栓的血管用人工辅助的方法单独切割出来, 然后用算法来自动寻找最佳分界点. 可以认为, 在分界点位置成的二类灰度集合的均方差是最小的, 这个分界点也就是血栓和血液之间的最优灰度分界点, 然后以这个分界点灰度为阈值进行二值化量化处理, 将灰度图象分成二类后, 以分类等到的血栓面积占切割区域的面积百分比做为衡量标准. 针对以上算法, 我们采用了Windows95平台下有Visual C++5.0编程语言设计了一个完整的图象处理平台, 并针对这个功能设计了一个单独的子程序, 添加了图象切割工具箱, 完成了多种图象切割功能; 自己设计完成了通用图象处理库, 设计了一个单独的血栓测量动态连接函数. 在图象处理过程中, 选用了五点十字中值滤波、低通Laplace滤波、图象直方图增强方法消除图象噪声. 然后利用上述算法寻找到了最佳分类点, 并成功地进行分类, 最后的测试结果以对话框报表形式输出. 针对多幅原始图象进行测试, 表明: 这种测量方法可以明显的降低噪声影响, 准确的切割图象, 寻找到最优分类点, 并成功的量化输出测试结果, 是一种有效的微白血栓测量方法.

(接上期) 3 X线图像中噪声的表现 X线图像中的噪声的表现,如图5所示.X线穿过密度均匀的被照体后,所成的影像的密度并不是处处均匀的,而是起伏波动的.