Nantong University

Hengrong Ju

Ming Li

Weiping Ding

Ying Sun

Yi Zhang

Haipeng Wang

Tao Hou

Yu Geng

Haowen Xue

Jiashuang Huang

Jinxin Cao

Xinjie Shen

Chun Cheng

Tingzhen Qin

Zhongyi Wang

Zhihao Feng

Jialu Ding

The present disclosure is a three-way U-Net method for accurately segmenting an uncertain boundary of a retinal blood vessel, includes: describing an uncertainty of a blood vessel boundary label, constructing an upper bound and a lower bound of the uncertain boundary based on the dilation operator and the erosion operator respectively to obtain a maximum value and a minimum value for the blood vessel boundary, and mapping the boundary with uncertain information into one range; combining an uncertainty representation of the boundary with a loss function, and designing a three-way loss function; training network parameters by adopting a stochastic gradient descent algorithm and utilizing a total loss of the three-way loss function; and designing and implements an auxiliary diagnosis application system for intelligently segmenting the retinal blood vessel with functions of the fundus data acquisition, the intelligent accurate segmentation and the auxiliary diagnosis for the retinal blood vessel.

Three-way U-Net method for accurately segmenting uncertain boundary of retinal blood vessel

A method for classifying multi-granularity breast cancer genes based on a double self-adaptive neighborhood radius includes large-scale gene locus data are read and normalized, and a data analysis is performed on the large-scale gene loci. An optimum value K is selected by adopting a combination of contour coefficients and a PCA dimensionality reduction visualization, and a model of information granulation is adjusted. A heuristic reduction algorithm is used to implement a multi-granularity attribute reduction of a self-adaptive neighborhood radius based on a cluster center distance and a multi-granularity attribute reduction of a neighborhood radius based on an attribute inclusion degree, and big data for breast cancer genes are classified and predicted by adopting a machine learning classification algorithm based on a SVM support vector machine.

Method for classifying multi-granularity breast cancer genes based on double self-adaptive neighborhood radius

Disclosed is a fully convolutional genetic neural network method for segmentation of infant brain record images. First, infant brain record image data is input and preprocessed, and genetic coding initialization is performed for parameters according to the length of a DMPGA-FCN network weight. Then, m individuals are randomly grouped into genetic native subpopulations and corresponding twin subpopulations are derived, where respective crossover probability and mutation probability pm of all the subpopulations are determined from disjoint intervals; and an optimal initialization value fa is searched for by using a genetic operator. Afterwards, fa is used as a forward propagation calculation parameter and a weighting operation is performed on the feature address featuremap. Finally, a pixel-by-pixel cross-entropy loss is calculated between predicted infant brain record images and standard segmented images to reversely update the weights, thus finally obtaining optimal weights of a network model for segmentation of the infant brain record images.

Fully convolutional genetic neural network method for segmentation of infant brain record images

Growing community of inventors

Jiangsu, China

Hengrong Ju