【作者】 李晓梅；

【导师】 郑成云；

【作者基本信息】 山东大学 ， 临床医学（专业学位）， 2014， 博士

【摘要】 目的红细胞刺激因子(Erythropoiesis-stimulating agents, ESAs)广泛应用于改善癌症患者伴发的贫血症状。尽管具有明显的有效性,最近研究表明,ESAs能导致严重的不良事件,甚至增加死亡率。该研究旨在系统评估ESAs用于癌性贫血患者中的利害关系,为临床合理安全使用ESAs提供理论依据。方法初步检索Medline, PubMed, Embase以及Cochrane数据库,共获取文献1569篇用于进一步的筛选。该meta分析共包含8项随机对照研究(n=2387),比较癌性贫血患者应用ESAs与安慰剂的差异。评估内容包括血红蛋白反应,红细胞输注频率及不良事件,例如深静脉血栓形成,高血压,研究期间死亡率及肿瘤进展。结果以合并比值比(OR)表示。用倒漏斗图表示发表偏倚。结果ESAs大大提高了血红蛋白浓度(OR=7.85,95%CI=5.85-10.53, p＜0.001),减少了红细胞输注次数(OR0.52,95%CI0.42to0.6, p＜0.001)。当目标血红蛋白数值不超过13g/dl寸,ESAs没有增加总的不良事件(OR=0.95, P=0.82),研究期间死亡率(OR=1.09,P=0.47)及深静脉血栓形成(OR=1.58,P=0.13)。结论ESAs能改善贫血,减少红细胞输注频率。当目标血红蛋白数值不超过13g/d1时,ESAs无增加癌性贫血患者血栓形成,死亡率及肿瘤进展的风险。目的获取血细胞的显微染色图像,利用图像处理和模式识别技术,对图像进行预处理及特征抽取,开发一套血细胞自动识别系统,为进一步研制高效、精确的血细胞检测分析及血液病理自动分析仪器奠定理论与实践基础,为血液病理分析自动化做出贡献。方法1)采集血细胞图像库；2)研究血细胞图片色彩空间模型,基于色彩空间转化及自适应二值化算法提出适用于血细胞图像自动分割检测的算法。该方法分割白细胞能有效的减弱光照对血细胞图像处理的影响；3)从纹理、形态、色彩三方面对分割后的白细胞图像进行特征参数联合提取；4)研究支持向量机算法在血细胞识别方面的应用,选择基于树的策略对白细胞进行多分类识别,并将细胞计数与血液病理分析相关联；5)对细胞自动识别系统进行matlab测试和验证。结果提出了高准确率、低时间开销的血细胞自动检测方法。结论完成了一套外周血细胞自动识别系统,测试系统软件运行正常稳定。更多还原

【Abstract】 Aim:Erythropoiesis-stimulating agents (ESAs) are widely used in the management of anemia in cancer patients. Despite their apparent effectiveness, recent studies have suggested that use of ESAs could result in serious adverse events and even higher mortality. The general aim of the current study was to provide valuable information guiding clinicians to apply ESAs for the patients in a good way and preventing from SEA-associated serious events by systemic evaluation of the benefits and risks of ESAs for cancer patients with anemia.Methods:The initial literature search covered Medline, PubMed, Embase, and the Cochrane Center Register of Controlled Trials, and identified1569papers that could be used as potential studies. The meta-analysis included eight randomized controlled trials (n=2387) comparing ESA versus placebo in cancer patients with anemia. The evaluation included hemoglobin (Hb) response, blood transfusion rate and adverse events, such as venous thromboemblism (VTE), hypertension, on-study mortality, and cancer progression. The results are expressed as pooled odds ratio (OR). Publication bias was assessed using funnel plot.Results:ESAs significantly increased the Hb concentration (OR7.85,95%CI5.85to10.53, p＜0.001), and reduced the red blood cell (RBC) transfusion rate (OR0.52,95%CI0.42to0.65, p＜0.001). When the target Hb was no more than13g/dl, ESAs did not increase the accumulated adverse events (OR0.95, P=0.82), the on-study mortality (OR1.09, P=0.47) and VTE (OR1.58, P=0.13). Conclusions:ESAs improves anemia and reduces the RBC transfusion frequency. ESAs are not associated with increased risks for VTE and cancer progression, and increased on-study mortality in cancer patients with anemia when the target Hb value is no more than13g/dl. Aim:Acquiring the microscopic images of blood cells, using image processing and pattern recognition technology, the images were preprocessed and their features were extracted, aiming to develop a set of blood cells automatic recognition system. Based on these works the efficient and accurate automatic analysis instrument of blood cells detection and hematopathology can be developed and manufactured. And these works will contribute directly to the hematopathology automatic analysis.Methods:1) Collecting blood cell image dataset;2) Analyzing blood cell image color space model, based on color space transformation and adaptive binarization algorithm, an automatic segmentation and detection algorithm for blood cells images were proposed. This method can weaken the effect of light on the blood cells image processing effectively when used to segment leukocytes images.3) The feature parameters were extracted from three aspects: texture, shape, and color, after segmenting leukocytes images.4) Studying the application of support vector machine (SVM) algorithm in blood cells recognition, multiple classification recognition were used on leukocytes images based on strategy tree, then the hematopathology analysis was associated with cell count.5) The cell automatic identification system were tested and verified in Matlab environment.Results:A high accuracy and low time cost method of automatically detecting the blood cells was proposed.Conclusion:A set of peripheral blood cell automatic identification system was completed. The software runs efficiently and stably.更多还原