摘要：【中文】本发明公开一种用于大规模脑病历分割的多粒度Spark超信任模糊方法，首先在Spark云平台上将大规模脑病历数据属性集分割至不同的多粒度进化子种群Granu‑population_i中；设计一种基于多粒度Spark超信任模型，构建多粒度种群内不同超级精英之间信任度；调整多粒度中心阈值，对超级精英使用多粒度子种群均衡调整策略进行动态更新，对大规模脑病历进行全局搜索分割与局部精化分割，超级精英在各自区域内能协同提取知识约简子集；最后求得大规模脑病历最优分割特征集并存储至Spark云平台中。本发明能稳定分割大规模脑病历知识约简集,为脑部疾病智能诊断和辅助治疗提供重要的诊断依据。 【EN】The invention discloses a multi-granularity Spark super-trust fuzzy method for large-scale brain medical record segmentation_iPerforming the following steps; designing a super trust model based on multi-granularity Spark, and constructing trust degrees among different super elite in a multi-granularity population; adjusting a multi-granularity central threshold, dynamically updating the super elite by using a multi-granularity sub-population balance adjustment strategy, and performing global search segmentation and local refinement segmentation on the large-scale brain medical record, wherein the super elite can cooperatively extract knowledge reduction subsets in respective areas; finally, the optimal segmentation feature set of the large-scale cerebral disease calendar is obtainedAnd storing the data into the Spark cloud platform. The invention can stably divide large-scale brain disease calendar knowledge reduction sets and provides important diagnosis basis for brain disease intelligent diagnosis and auxiliary treatment.

摘要：【中文】本发明涉及到医学信息智能处理领域，具体来说涉及一种用于视网膜病变分类的卷积神经网络权值优化方法。该方法首先获取眼底图像训练集、及其对应的多病变标签；通过单种群蛙跳算法寻找最优初始权值，然后构建卷积神经网络中的卷积层、池化层和全连接层，将最优初始权值作为第一次前向传播计算的参数；将视网膜中四种病变的四个预测值分别与真实值进行交叉熵损失计算并求和得到损失值，判断损失值是否异常，如果异常则围绕前一次前向传播的权值生成蛙群，寻找最优蛙更新网络权值；否则采用梯度下降算法更新网络权值；最后对最终权值进行优化。本发明能有效提高眼底图像多病变检测的准确率，对视网膜疾病和辅助治疗具有较强应用价值。 【EN】The invention relates to the field of medical information intelligent processing, in particular to a convolutional neural network weight optimization method for retinopathy classification. Firstly, acquiring a fundus image training set and a multi-lesion label corresponding to the fundus image training set; searching an optimal initial weight through a single swarm leaping algorithm, then constructing a convolution layer, a pooling layer and a full-link layer in a convolutional neural network, and taking the optimal initial weight as a parameter for the first forward propagation calculation; respectively carrying out cross entropy loss calculation on four predicted values of four pathological changes in retina and a true value, summing to obtain a loss value, judging whether the loss value is abnormal, if so, generating a frog group around a previous forward propagation weight, and searching for an optimal frog updating network weight; otherwise, updating the network weight by adopting a gradient descent algorithm; and finally optimizing the final weight. The invention can effectively improve the accuracy of fundus image multi-lesion detection and has stronger application value to retinal diseases and adjuvant therapy.

摘要：【中文】本发明涉及到眼底血管图像聚类操作技术领域，具体来说涉及一种用于眼底血管图像聚类分割的近似骨架蛙群编号方法。本发明借助聚类方法，对眼底图像进行分割处理，根据病变点高亮的特性对病变点进行定位和剔除。为了获得更好的聚类分割效果，采用智能算法中较为有效且便于理解的混合蛙跳算法对K‑means算法进行改进并使用近似骨架进一步充分利用算法获得的局部最优解，改进后的算法能有效克服原始K‑means算法易于收敛至局部最优而无法有效进行图像分割缺点，获得更好的眼底血管聚类分割效果，更准确的分离出眼底血管的病变点。 【EN】The invention relates to the technical field of fundus blood vessel image clustering operation, in particular to an approximate skeleton frog cluster numbering method for fundus blood vessel image clustering segmentation. The invention carries out segmentation processing on the fundus image by means of a clustering method, and positions and eliminates lesion points according to the characteristic of highlight of the lesion points. In order to obtain a better clustering segmentation effect, a mixed frog-leaping algorithm which is effective and convenient to understand in an intelligent algorithm is adopted to improve a K-means algorithm, an approximate skeleton is used for further and fully utilizing a local optimal solution obtained by the algorithm, and the improved algorithm can effectively overcome the defect that the original K-means algorithm is easy to converge to local optimal and cannot effectively perform image segmentation, so that a better fundus blood vessel clustering segmentation effect is obtained, and lesion points of fundus blood vessels are more accurately separated.

摘要：【中文】本发明公开一种用于微血管瘤病历图像的超平面近邻分类方法。该方法首先对糖尿病性眼底图像数据进行预处理和分割操作，从处理后的眼底病历图像中提取出微血管瘤病历图像的病变区域；接着将微血管瘤病变的图像区域形态学特征、纹理特征及灰度特征转化为l维数据向量x_i；然后将数据分为训练数据X_tr和测试数据X_te，通过对训练数据X_tr进行训练得到一个包括分类超平面Hyper、支持向量集合X_sv、距离阈值t、最近邻居个数k、和谱哈希编码码长nb的高效分类模型；最后测试数据X_te预测时依据测试样本到分类超平面Hyper的距离与距离阈值t的关系，分别采用支持向量机模型和融合谱哈希算法的近邻算法进行预测，并综合相关预测结果。本发明能对提取出的眼底病历中微血管瘤病历图像特征进行快速有效分类，具有较高的分类准确率，大大降低了微血管瘤病历图像特征分类的执行时间。 【EN】The invention discloses a hyperplane nearest neighbor classification method for a microangioma medical record image. Firstly, preprocessing and segmenting diabetic fundus image data, and extracting a lesion area of a microangioma medical record image from a processed fundus medical record image; then, the morphological characteristics, the textural characteristics and the gray-scale characteristics of the image area of the microangioma lesion are converted into a data vector x with the dimension of l_i(ii) a The data is then divided into training data X_trAnd test data X_teBy comparing training data X_trTraining to obtain a set X comprising classified hyperplane and support vector_svThe efficient classification model comprises a distance threshold t, the number k of nearest neighbors and the length nb of a spectral Hash coding code; final test data X_teAnd during prediction, according to the relation between the distance from the test sample to the classified hyperplane and the distance threshold t, respectively adopting a support vector machine model and a neighbor algorithm of a fusion spectrum hash algorithm to perform prediction, and integrating related prediction results. The invention can extract the micro-particles in the fundus medical recordThe hemangioma medical record image features are quickly and effectively classified, the classification accuracy is high, and the execution time of the microangioma medical record image feature classification is greatly reduced.