北京生物医学工程

结合二维灰度直方图与PCNN的医学图像多类分割

Multi-segmentation of Medical Image Based on 2D Gray Histogram and Pulse Coupled Neural Networks

作者：钟瑾施俊常谦

单位：上海大学通信与信息工程学院（上海200072）

分类号：

出版年·卷·期（页码）：2011·30·1（26-31）

摘要：

针对医学图像中对组织器官多类分割的要求，提出一种结合二维灰度直方图的脉冲耦合神经网络（pulse coupled neural networks，PCNN）图像多类分割算法。首先根据PCNN模型的局部连接作用和阈值衰减特性对图像进行多类分割，然后利用基于类内最小离散度的二维直方图算法计算出PCNN网络迭代时的最佳门限值，从而实现医学图像的多类分割。通过对仿真的正常颅脑和非正常的颅脑核磁共振图像进行测试，结果显示本PCNN图像多类分割算法能够有效地分割出核磁共振颅脑图像中不同脑组织。因此，本文算法具有应用于医学图像的多类分割的可行性，并提高计算机辅助分割医学图像的准确性。

An image multi-segmentation algorithm was proposed to segment different tissues in medical image. This algorithm was based on the 2D gray histogram and pulse coupled neural networks (PCNN). The features of local connection and threshold attenuation of PCNN model were used to segment different tissues. The 2D gray histogram with minimum dispersion within cluster was then used to obtain the optimal threshold in PCNN iteration. The simulated normal and abnormal magnetic resonance brain images were tested for this algorithm. The results indicate that the proposed algorithm can effectively segment different tissues in brain images, which suggests that this algorithm has the feasibility to be used for medical image multi-segmentation, and can improve the accuracy of computer-aided medical image segmentation.

参考文献：

［1］边肇祺. 模式识别［M］.第2版.北京：清华大学出版社, 2000.
［2］Eckhorn R, Reitboeck H J, Arndt M. Feature linking via synchronization among distributed assemblies: simulation of results from cat cortex［J］. Neural Computer, 1990, 2(3): 293-307.
［3］Eckhorn R, Frien A, Bauer R. High frequency oscillations in primary visual cortex of awake monkey［J］.Neuroreport, 1993, 4(3): 243-246.
［4］刘潇,孔祥维. 基于直方图的脉冲耦合神经网络(PCNN)多门限图像分割方法［J］. 佳木斯大学学报, 2004, 22(1): 38-44.
［5］Brink AD. Thresholding of digital images using Two-dimensional Entropies ［J］. Pattern Recognition, 1992, 25(8): 803-808.
［6］吴一全,张金矿. 基于改进的二维最大熵及粒子群递推的图像分割［J］.计算机辅助设计与图形学学报, 2008, 20(10): 1338-1344.
［7］郭海涛,王连玉,田坦. 利用二维属性直方图的Otsu的自动阈值分割方法［J］. 光子·激光, 2005, 16(6): 739-742.
［8］周云燕,杨坤涛,黄鹰. 基于最小类内离散度的改进Otsu分割方法的研究［J］. 华中科技大学学报, 2007, 35(2): 101-103.
［9］BIC McConneu Brain Imaging Center ［EB/OL］.http://noodles.bic.mni.mcgill.ca/Main/HomePage.
［10］Zhang Y, Sankar R, Qian W. Boundary Delineation in Transrectal Ultrasound Image for Prostate Cancer［J］. Comput Biol Med, 2007, 37: 1591-1599.
［11］Rivera M, Ocegueda O, Marroquin J L. Entropy-controlled quadratic markov measure field models for efficient image segmentation［J］. IEEE Trans Image Process, 2007, 16: 3047-3057.

服务与反馈：

【文章下载】【加入收藏】

提示：您还未登录，请登录！点此登录