人脑CT病变区图像准确诊断仿真研究

摘要

An accurate diagnosis method of human brain disease via modified method of image segmentation is proposed.The FCM cluster principle is used to carry out original segmentation for interesting region of image of the lesion area and the outline of the interesting region of image is extracted,then the information of multi-scale gradient vector is acquired to build energy function with boundary type.The Hessian matrix is used to carry out enhancement filter for three-dimensional (3D) volume data of the interesting region and the network planning for 3D CT image after enhancement filter integrated with gray level and space information is built.Moreover,integrated with graph cut theory,the segmentation of network planning is achieved.Thus,the image segmentation of lesion area is completed.Experimental results show that the method can effectively improve accuracy of disease diagnosis of human brain.%通过对人脑CT病变区图像进行分割从而实现人脑疾病的准确诊断,可以有效降低人脑疾病导致的死亡率.对人脑CT病变区图像的分割,需要获取多尺度梯度矢量信息,利用改信息构造图像区域边界型能量函数,完成对图像的精准分割.传统方法先计算出CT图像符号距离函数,并结合快速步进法生成符号表,但忽略了构造图像区域边界型能量函数,导致分割精度偏低.提出通过图像分割改进方法实现人脑疾病准确诊断.利用FCM聚类原理对人脑CT病变区图像感兴趣区域进行初分割,提取CT图像感兴趣区域轮廓,获取多尺度梯度矢量信息构造边界型能量函数,利用Hessian矩阵对人脑感兴趣区域3维体数据增强滤波,对增强滤波后的三维CT图像,结合灰度和空间信息构造网络图,结合图割理论实现网络图分割,由此完成对人脑CT病变区图像分割.实验结果表明,所提方法能够在有效提升人脑疾病诊断正确率.