能谱CT优化胃肿瘤扫描辐射剂量对肾上腺嗜铬细胞瘤的诊断价值

目的:探讨能谱CT优化胃肿瘤扫描辐射剂量对肾上腺嗜铬细胞瘤的诊断价值。方法:采用回顾性、抽样、随机研究方法选择2012年9月到2017年2月在我院诊治的肾上腺嗜铬细胞瘤患者59例作为研究对象,所有患者都给予常规CT扫描与能谱CT优化胃肿瘤扫描,记录和比较辐射剂量与图像质量。结果:所有病例包膜均完整,边缘清楚,肿瘤内见单发或多发低密度区,肿瘤实质区呈不均匀显著强化。常规CT与能谱CT的图像质量主观评分分别为3.89±0.45分和4.54±0.34分;常规CT与能谱CT图像的胃肿瘤CT值分别为31.94±6.39HU和35.29±5.19HU,对比都有显著差异(P0.05)。能谱CT图像的膀胱和皮下脂肪图像噪声值都显著低于常规CT图像,对比差异都有统计学意义(P0.05);能谱CT扫描的CTDIvol和DLP分别为12.39±3.48mGy和624.10±39.19mGy.cm,都显著低于常规CT扫描的14.09±4.13mGy和653.92±56.29mGy.cm(P0.05)。结论:能谱CT优化胃肿瘤扫描在肾上腺嗜铬细胞瘤诊断中的应用能有效减少辐射剂量与图像噪声,提高图像CT值与主观质量,临床应用价值更高。     

用家用数码照相机直接经显微镜目镜摄取高质量的数码图像

在生物学和医学的科研、教学中，往往需要使用生物显微镜和解剖镜。而与之相配套的照相装置，是保存、发表和教学展示光学显微图像的必备设备。一般的显微照相装置需要配有专用的光学适配装置，用于特定型号的显微镜，拥有成套设备的成本较高。本文探讨了使用普通家用数码照相机直接经生物显微镜目镜，摄取高质量、高分辨率数码光学显微图像的可行性和影响因素。     

人工智能大模型医学应用研究

近年来,以自然语言处理和视频图像分析为主的人工智能大模型技术得到快速发展,其基本特征是聚焦相关应用领域的共性需求,通过大数据、强算力和复杂算法的高效协同与深度融合,构建通用预训练模型,广泛适配下游任务,有力提高模型的处理性能与研发效率.因此,大模型技术为医学人工智能高质量发展提供了难得契机.本文通过全面梳理国内外大模型的研究进展、关键技术与核心算法,分析总结生物医学领域一系列标准数据集和预训练模型的发展特点,结合医学人工智能的研发实践,深入剖析医学领域大模型构建的应用需求、解决思路与研发经验,助力推动医学大模型创新发展.     

菌丝形态分化与头孢菌素C合成的关系*

利用显微图像分析法对顶头孢霉菌的菌丝形态进行了定量研究,并统计分析了头孢菌素C发酵过程中的菌丝形态的变化规律,具体对菌丝长度、菌丝宽度和菌丝生长单位进行了定量分析,分析了菌丝形态分化与头孢菌素C合成的关系。研究表明头孢菌素C的合成是在细长菌丝分化成膨大菌丝片段后才启动的,头孢菌素C可能主要是在膨大菌丝分化成节孢子的过程中被合成。     

中国人胃癌染色体17q21区域基因遗传不稳定性研究

To investigate microsatellite instability (MSI) and loss of heterozygosity (LOH) of locus D17S396, D17S579 and D17S855, and their effect on the expression of nm23H1 and BRCA1 of gastric cancer, which would provide experimental basis for clinical treatment and prognosis analysis of gastric cancer. DNA was extracted from paraffin-embedded materials. Polymerase chain reaction-single strand conformation polymorphism (PCR-SSCP) was used to analyze MSI and LOH. Expression of nm23H, and BRCA1 was detected by Envision immuno-histochemistry and Leica-Qwin computer imaging techniques. In the forty cases of gastric cancer, the frequency of MSI, LOH and nm23H1 protein were 20.00%, 17.50% and 55.00% respectively at locus D17S396, while at locus D17S579, the frequency of MSI, LOH and BRCA1 protein were 22.50%, 15.00% and 37.50% respectively; at locus D17S855, the frequency of MSI, LOH and BRCA1 of thirty-seven cases were 18.92%, 18.92%, 37.84% respectively. In tumor node metastasis (TNM) staging, at locus D17S396, D17S579 and D17S855, MSI in stages I + II appeared more frequently than that in stages III + IV, while LOH appeared the contrary tendency. In the group of metastasis of gastric cancer, MSI had a less frequency (5.00%) than that with no metastasis (35.00%, P < 0.05) at locus D17S396, but LOH appeared more frequently (30.00%) than that with no metastasis (5.00%, P < 0.05). At locus D17S579, MSI had an increasing tendency with the degree of tumor differentiation (50.00% in high differentiation cases, 20.00% in middle differentiation cases, and 0% in low differentiation cases, P < 0.05). The frequency of nm23H1 and BRCA1 protein in stages TNM I + II was higher than that in stages TNM III + IV; and that in higher differentiation cases was higher than in poor differentiation cases. The frequency of nm23H1 protein in the group of metastasis (30.00%) was less than that with no metastasis significantly (80.00%, P<0.01). The frequency of nm23H1 protein in the group positive to MSI (87.50%) was higher than that in the group negative to MSI (46.88%, P < 0.05). However, nm23H1 protein in group positive to LOH (14.29%) was lower than that in the group negative to LOH (63.64%, P < 0.05). The frequency of BRCA1 protein in the group positive to MSI (66.67%) was more than that in the group negative to MSI (29.03%, P < 0.05). The results of experiments indicate that MSI and LOH may separately control the development of sporadic colon cancer with different pathways. MSI may be an early period molecule marker for sporadic colon cancer, enhanced expression of nm23H1 protein can effectively inhibit colon cancer metastasis and improve prognosis of sporadic colon cancer patients. By comparison, LOH mostly arises in the late period of sporadic colon cancer and endows a high aggressive and poor prognostic phenotype. nm23H1 protein could effectively restrain gastric cancer metastasis and development; and BRCA1 protein could restain tumor from becoming lower differentiation.     

Takayama pulchellum原位增长率测定的研究

现今用于测定有害藻华生物原位增长率的方法很难在现场进行有效地应用,因此我们尝试采用基于rRNA、叶绿素和总蛋白质浓度的定量测定方法来估算其细胞增长率.本研究重点对Takayama pulchellum细胞株(TPXM)的rRNA含量和特定生长率的关系进行了探索.用荧光标记的肽核酸探针结合全细胞杂交方法,结合流式细胞仪和专业图像分析软件对T.pulchellum单细胞水平的rRNA进行了定量检测.结果表明不同生长阶段的T.pulchellum单细胞rRNA水平与其相应的细胞特定生长率有较好的相关关系(R2=0.7293,p＜0.01,n=14),单细胞蛋白质含量、rRNA水平与同步后的细胞周期变化情况也有较好的相关关系(R2＝0.6984,p＜0.01,n=14).上述结果提示单细胞rRNA含量可作为指示细胞周期变化和细胞特定生长率的较好指标.     

数字照片在树干解析上的应用

介绍了一种利用摄影测量技术开展树干解析的新方法.通过求算横截面积,换算成理论直径及年轮宽度.此法可以提高测量精度,特别是在树干横截面不规则的情况下,可以提高准确度.选择一个基本圆满的圆盘,直接测量直径和由面积换算比较年轮宽度,二者差异平均为10%,最大达32%.由于图像上直接测量的是面积,树皮的横截面可以更加准确获得,而传统的方法无法克服这个缺点.数字照片还是储存年轮信息的理想形式,便于保存和复查.本法使用的是普通的数码照片,方便适用,是开展森林精密测量简单易行手段.     

昆虫图像自动鉴别技术

昆虫是地球上物种多样性最为丰富的生物类群,其物种鉴定任务复杂而艰巨,可靠的物种鉴定是开展昆虫学工作的重要基础之一。当前,国内外的人工昆虫物种鉴定能力均不能满足实际需求,因而人们开始不断探索利用计算机自动鉴定昆虫的原理和方法。目前,模式识别技术的迅猛发展已为昆虫图像的自动鉴定提供可能。文章概述昆虫图像自动鉴定技术研究的历史与现状,总结主要原理和方法,介绍工作流程,并展望发展前景。     

基底周期拉伸引起ECV-304细胞形态学变化的分析

对ECV-304细胞施以最大应变10％、0.67Hz的周期拉伸,利用计算机图像处理系统,对周期拉伸过程中ECV-304细胞的取向调整进行了形态学分析,结果显示：周期拉伸能引起细胞长轴取向垂直于最大主应变方向;加载12h内细胞长短径比增加,12-24h之间长短径比下降,随后趋于稳定;细胞在周期拉伸最大主应变方向上的最大截距缩短,而在垂直于最大主应变方向上的最大截距延长;取向调整的过程与长短径比增大的过程有显著的相关性。表明在周期拉伸过程中的取向调整是一个细胞具有方向差异性的变形过程,而不是刚性的旋转或位移。     

共聚焦激光检眼镜下眼底图像的非特征配准约束模型研究

研究共聚焦激光检眼镜下不基于特征提取的眼底图像自动配准方法中的运动约束模型,从成像机理上分析共聚焦激光检眼镜下图像对间的运动模式,并分析比较多种实际全局运动模型约束下的配准精度和效率,进而给出一种由粗到细的复合约束模型对眼底图像进行配准。实验结果证实了该模型效果良好。