一种半监督的彩色图像分割方法

提出一种基于半监督EM聚类的彩色图像分割方法, 算法利用了有限的人工信息, 即在图像上点击有限的几个点以标识对应区域之间的关系, 从而得到满足给定限制的精确图像分割结果。算法首先对图像进行量化处理, 而后在量化后的色彩空间中集成先验的分割信息进行色彩聚类。实验表明算法运行速度快, 分割效果好, 具有很高的应用价值。     

模糊集理论在图像分割中的应用综述

图像分割是许多图像分析和图像处理过程中的重要步骤。由于人的视觉特性和数字图像本身所具有的模糊性,使得图像分割问题是典型的结构不良问题,由Zadeh提出的模糊集合论具有描述不良问题的能力,模糊集合论应用于图像分割是针对图像模糊性非常有效的方法。目前模糊集理论在图像分割中的应用有许多成功的方法。本文就基于模糊理论的图像分割方法进行了综述,主要归纳了模糊阈值分割、模糊聚类分割以及模糊神经网络分割等方法并简要讨论了各种方法的特点。     

一种基于高斯混合模型与Markov建模的灰度图像分割方法

当灰度图像中存在区域间灰度变化不明显或者含有噪声时,图像分割效果会受到比较严重的影响,本文针对此类问题,基于高斯混合模型,提出了一种改进的灰度图像分割算法。首先,基于马尔可夫随机场建模,将梯度因素引入邻域约束,推导图像的能量函数。然后,采用改进的期望最大（EM）算法对能量函数进行求解,E步通过图割法求解各像素点的分类,M步通过改进的期望最大（EM）算法求解高斯混合模型中的各参数。实验结果表明,本文的方法相对于直接用图割法能够求得更低的能量值,获得较好的分割结果。     

皮肤黑素细胞肿瘤图像综合分割方法研究

目的 本文针对黑素细胞肿瘤（Melanocytic Tumor MT）图像情况复杂，较难分割的问题，提出了一种综合数字图像分割算法，探讨MT的早期诊断。方法 首先应用统计区域融合方法（SRM）实现图像分割成多块纹理一致的区域。然后对图像以HSV彩色空间的H和S分量为特征，使用K均值聚类算法将图像聚为9类。最后，将聚类结果在HSV彩色空间的H和S分量值分别映射到[0，1]区间，再分别对H分量和S分量取阈值，得到最终的边界分割结果。结果 对MT图像能够按照其纹理差异将其有效划分为多个区域，较为准确标识出皮损区域。结论 综合对多种方法结果的对比，本方法优于传统的大津阈值法、K均值法和活动轮廓法。同时对过去基于SRM的MT图像分割方法进行了改进，在处理复杂MT图像时效果明显好于传统方法。     

基于分解的二维指数灰度熵图像阈值分割

现有最大Shannon熵或Tsallis熵阈值选取方法仅仅依赖于图像灰度直方图的概率信息,而没有直接考虑类内灰度的均匀性;且用对数定义的Shannon熵还存在无定义值和零值的问题。为此,提出了一维和基于分解的二维指数灰度熵阈值分割方法。首先给出了指数灰度熵的定义及其一维阈值选取方法;然后将其推广得到二维指数灰度熵阈值选取公式,提出了基于分解的二维指数灰度熵阈值分割方法。通过分别求原像素灰度级图像和邻域平均灰度级图像的一维指数灰度熵最佳阈值,并将其组合求解二维指数灰度熵最佳阈值,大大缩小了搜索空间,使计算复杂度由O（L4）降为O（L）。实验结果表明,与基于基本粒子群算法的二维最大Shannon熵法、二维最大Tsallis熵法以及采用递推的二维交叉熵法相比,所提出方法的分割效果具有明显的优势,且运行时间大幅减少。     

基于色度学准则分析的免疫组化彩色图像C-均值聚类分割技术研究

免疫组化彩色图像分割在免疫组化染色定量分析中有重要的应用价值。C-均值聚类算法(CMA)是根据一定的相似性准则将图像分成C类。由于运算十分耗时,直接限制了C-均值聚类算法在彩色图像分割领域的应用。本文针对免疫组化彩色图像特点,提出了分割免疫组化彩色图像的色度学准则,即:用每个像素的R分量减去B分量,根据其差值是否大于0将相应的像素分为两大类:(R-B)≥0类和(R-B)<0类,自动分离出图像的阳性细胞区域和阴性细胞区域。在此基础上我们改进了C-均值聚类分割的方法:①针对上述两大类图像的色彩分布的特点确定初始聚类中心;②分别对上述两大类像素在一个色度学空间聚类;③根据每次迭代过程中聚类中心的变化趋势来预测下一次迭代后可能的聚类中心,从而减少迭代的次数。改进之后的C-均值聚类分割方法减少了聚类的样本数量,降低了算法的复杂度,且由于迭代次数的减少,运行速度得到了提高。实验结果表明,本文建立的技术方法能有效地分割免疫组化彩色图像。     

基于改进自生成神经网络的皮肤镜黑色素细胞瘤图像分割

针对皮肤镜黑色素细胞瘤图像,提出一种基于自生成神经网络(self-generating neural net-work,SGNN)的自动分割算法。算法首先采用区域生长的方法将图像进行粗分割;然后将每一个子区域看作一个叶节点,根据节点之间的相邻关系定义连接规则;最后采用SGNN对这些节点进行聚类,完成黑色素细胞瘤图像的分割。本文方法克服了传统SGNN算法对样本训练顺序敏感的缺陷,提高了效率,实验结果表明,该方法能够自适应确定聚类数目并准确分割黑色素细胞瘤图像。     

基于特征聚类的多Atlas分割研究

目的研究基于聚类的多Atlas分割方法对正常组织感兴趣区分割的改善, 以达到更好的危及器官的勾画效果。方法选取2019—2020年浙江省肿瘤医院已完成治疗的100例宫颈癌患者的CT图像作为Atlas图谱库。按照危及器官(膀胱、直肠和外轮廓)的体积特征参数作为测度, 利用k均值聚类(k-means)算法将Atlas图谱库划分成若干子集。将待分割图像匹配到相对应的图谱库中进行多Atlas分割。使用相似性系数(DSC)对分割结果进行评价分析。结果以30例患者作为测试组, 比较了不同聚类方法所生成的子图谱库对图像分割结果的改进。相较于一般多Atlas分割, 聚类多Atlas分割方法能显著提高膀胱(DSC为0.83±0.09∶0.69±0.15, P<0.001和直肠(DSC为0.7±0.07∶0.56±0.16, P<0.001)的分割准确性, 但左、右双侧股骨头(0.92±0.04、0.91±0.02)和骨髓(0.91±0.06)的差异无统计学意义。并且聚类多Atlas分割方法平均分割时间短于一般多Atlas分割方法(2.7∶6.3 min)。结论聚类多Atlas分割方法不但会减...     

基于形态特征的淋巴瘤病理图像分级分割方法

目的: 提出一种淋巴瘤病理图像分级分割方法。方法: 针对淋巴瘤病理图像中组织灰度在RGB颜色通道中的分布形态特点, 采用图像代数运算增大淋巴瘤病理图像中不同染色质组织之间的灰度差异, 采用最大类别方差方法, 建立一个融合多通道灰度分布形态信息的分割模型,分离出淋巴瘤病理图像中不同染色质的区域; 再利用区域的面积、灰度纹理粗糙度、圆形度等几何形态特征建立二级分割模型, 分离出同染色质中不同的组织靶区。结果: 实现了淋巴瘤病理图像的自动分割。结论: 试验结果表明该方法速度快, 分割效果理想, 为淋巴瘤病理图像的定量分析打下良好的基础。     

一种结合空间信息的图像阈值分割新方法

针对基于灰度分布的图像阈值分割方法的不足,本文通过结合使用灰度级和空间相关信息将分类像素的特征空间从一维推广到二维,利用二维Otsu方法得到稳健的初始分割,然后使用松弛迭代的区域增长技术来判决边缘像素的归属。实验结果表明,本方法能够获得更好的阈值分割效果和更强的鲁棒性,并且大大降低了时间复杂度。     

Measuring information strategies in oncology - developing an information satisfaction questionnaire

The benefits of providing accurate and culturally sensitive information for patients are now undisputable and information strategies are generally being adopted nationally. Measuring their benefits and adapting strategies to local needs is now a new challenge. Although questionnaires to measure global satisfaction are available, no concise tool has previously existed to specifically measure the information aspects of care. The aims of this study are: to establish which aspects of information were most important to the patients during their cancer pathway; to develop a robust single-page questionnaire that is sensitive to patient priorities to measure and compare satisfaction between patient groups; and to audit information provision within routine clinical practice. Seventy-seven consecutive patients were asked to complete a priority list of which aspects of information were most important to them during their recent cancer care. A total of 303 answers were categorized into five final categories of equal weighting. The data were combined with demographic issues to produce a robust one-page questionnaire which has been adopted formally by the West Anglia Cancer Network. One hundred consecutive patients were then audited to highlight the information shortcomings of a busy community oncology department and the findings are reforming information services. We recommend the use of this practical audit tool in routine clinical practice, which can be downloaded free from cancernet.co.uk in seven languages.     

Family members' experiences, information needs and information seeking in relation to living with a patient with oesophageal cancer

A qualitative study was conducted with the aim to describe family members' experiences, information needs and information seeking in relation to living with a patient suffering from oesophageal cancer. Data were collected by means of semi-structured interviews with nine family members. A content analysis was used in order to organize data. It was found that family members were not aware of the severe diagnosis and the illness caused intrusions on the family. The time following diagnosis family members' information seeking was low. They used interpersonal as well as mass media sources to obtain knowledge about and handle the uncertainty related to the illness. Some family members did not actively seek information. In conclusion, healthcare professionals are expected to improve quality of cancer care. These improvements are unlikely to occur without an understanding of family members' needs. This study shows that the family members were unprepared of receiving a diagnosis of oesophageal cancer. They emphasized the importance of including the children in the care given. Moreover, the whole family was faced with uncertainty, which led to intrusion on everyday life. Therefore, family members used different strategies for managing the uncertainty, whereof one was searching for information. The primary source of information was the physician.     

The experience and impact of contacting a cancer information service

Our objective was to review the quality of service delivered by a national cancer information service–the British Association of Cancer United Patients (BACUP)–and to consider the implications for future policy and practice. We used a sample of callers over a 10-day period, responding to a structured postal questionnaire. The sample included patients, relatives, friends, and general public who called the service with any concern relating to cancer. Of the 406 invited to take part, 282 responded (69%). The questionnaire was designed to evaluate the callers' perceptions of the information they received, their perceptions of the nurse's communication skills, the emotional impact of the call, and their overall satisfaction with the service.
The results indicate that over 90% of callers evaluated the information they received and the nurse's communication skills positively, and there was no difference between the two major groups of callers, i.e. patients and relatives/friends. Both groups reported that the call had a positive emotional impact on them and that they were satisfied with the service. Emotional impact was predicted by quality of information for patients and nurse's communication skills for friends/relatives; overall satisfaction was predicted by nurse's communication skills for patients and both measures for relatives/friends. We conclude that the telephone is an effective medium for conveying information about cancer, and BACUP is meeting its needs. The needs of patients and relatives/friends are different–principally information for patients and support for relatives/friends–and it may be possible to improve training still further by focusing on those differences. The main challenge for cancer information services is to meet the ever-increasing demand.     

Unpacking cancer patients' preferences for information about their care

Objective: Patient-centered decision making requires cancer patients be actively involved and feel sufficiently informed about their care, but patients' preferences for information are often unrecognized or unmet by their oncologist, particularly for more distressing topics. This study examined cancer patients' preferences for information about three care-related topics: (1) diagnostic information, (2) treatment costs, and (3) prognosis. We tested whether factors known to influence information preferences (psychological distress, control preferences, and financial distress) were differently associated with information preferences for each topic. Methods: Cancer patients (N = 176) receiving ongoing treatment completed a questionnaire that assessed their out-of-pocket treatment costs, psychological distress, preferences for control over their medical decisions, and the amount of information they desired and received from their oncologists about the three topics. Results: Patients' preferences were less often met for treatment cost information than for the other topics, p < 0.001, with half wanting more cost information than they received. One-third of patients also wanted more prognostic information than they received. Patients' preferences for diagnostic information did not differ as a function of financial burden, distress, or control preferences, ps > 0.05. Preferences for cost information were greater among patients who preferred more control over their medical decisions, p = 0.016. Patients' preferences for prognostic information were greater among those desiring more control and with lower distress, ps < 0.05. Financial burden was not associated with information preferences. Conclusion: Appreciating the variability in information preferences across topics and patients may aid efforts to meet patients' information needs and improve outcomes.     

Mapping the information resources available to patients with colorectal cancer

KING S.J., LIVINGSTON P.M., TURNER L., BYRNE K., JOHN M., SIDWELL J. & SCOTT C. (2010) European Journal of Cancer Care
Mapping the information resources available to patients with colorectal cancer The objectives of this study were to identify gaps in information provision along the colorectal cancer (CRC) treatment pathway as provided by health services within the North Eastern Metropolitan Integrated Cancer Service in Victoria Australia; to evaluate the information and recommend consistent, high quality health information resources; and to recommend strategies to improve delivery of patient information. A random sample of health professionals (n= 47) from various disciplines at eight health service sites participated in semi‐structured interviews regarding the types of information they provided to CRC patients. Information items were mapped against a published CRC patient management framework and evaluated. A total of 193 information items were collected with 24 items specific to CRC. Gaps in information provision were evident in the community, at diagnosis, in clinics, when treatment was determined and when completed. The quality of information delivery to CRC patients across the public health sites was variable. Resources were often unavailable, out of date and inaccessible in other languages. Results indicate a need to improve health information availability and resource delivery to all CRC patients across different health services particularly at diagnosis and after treatment. Further research is required to determine patient preferences for information about CRC.     

Patient information preferences among breast and prostate cancer patients

Preferences for information about their disease and treatment were collected from 392 patients who had been treated for either breast or prostate cancer an average of 2 years previously. Type of information that they had received, their ratings of its value to them and their preferred format for further information were examined. The most common and most preferred form of information was through doctor interview, followed by educational booklets. Prostate cancer patients preferred videotapes; breast cancer patients tended to prefer individualized approaches including a tour of the department. Effects of age, educational levels, occupational backgrounds and self‐reports for anxiety and depression at the time of the survey and at time of diagnosis were analysed. Prostate cancer patients who were most severely depressed showed a preference for not receiving any information at all, perhaps reflecting a tendency towards withdrawal.     

放疗信息管理系统的研发与应用

目的 研发放疗信息管理系统。方法 放疗信息管理系统采用B/S架构模式, 以ACCESS 2010为前台数据库服务器运行在医院局域网络上, 后台数据库为调用Oncentra TPS和Mosaiq治疗系统中SQL Server2008。使用ASP网络编程语言编写, 整个系统是在Macromedia Dreamweaver 8平台开发。基于互联网信息服务6.0(IIS 6.0)服务组件构建系统服务器, 用IE浏览器实现客户端访问服务器功能。结果 该信息系统共包括系统设置模块、物理师模块、医生模块、技师模块、数据统计与分析、数据下载、视频教学7个模块。结论 放疗信息管理系统操作简单, 实时性强, 数据安全, 运行稳定, 是放疗资源高效利用重点建设和放疗信息化建设的重要组成部分。     

病理信息管理系统的建设与应用

目的建设一个数字化、网络化、可质控的病理信息管理系统工作平台,与医院共享患者所有诊疗信息,为病理诊断、科研和教学服务。方法根据病理科实际工作流程诸多环节,通过英飞达软件（上海）有限公司提供的技术支持,依托Oracle数据库及VB.NET开发软件,建设网络版的新型实用的病理信息管理系统。结果该系统建成投入运行后,病理检查的流程更为明晰,质控更加精确,实现了新旧病理信息、临床检查及影像信息、临床医嘱等相关信息同界面调用显示。结论新构建的病理信息管理系统适应国内三级医院病理科的实际工作需要,优化了病理检查的工作流程,实现了临床与病理全信息交互。     

Information needs and sources of information for patients during cancer follow-up

Background

Now more than ever, cancer patients want health information. Little has been published to characterize the information needs and preferred sources of that information for patients who have completed cancer treatment.

Methods

We used a nationally validated instrument to prospectively survey patients attending a cancer clinic for a post-treatment follow-up visit. All patients who came to the designated clinics between December 2011 and June 2012 were approached (N = 648), and information was collected only from those who agreed to proceed.

Results

The 411 patients who completed the instrument included individuals with a wide range of primary malignancies. Their doctor or health professional was overwhelmingly the most trusted source of cancer information, followed by the Internet, family, and friends. The least trusted sources of information included radio, newspaper, and television. Patients most preferred to receive personalized written information from their health care provider.

Conclusions

Cancer survivors are keenly interested in receiving information about cancer, despite having undergone or finished active therapy. The data indicate that, for patients, their health care provider is the most trusted source of cancer information. Cancer providers should ask patients about the information they want and should direct them to trusted sources.