基于改进型脉冲耦合神经网络的乳腺图像肿块分割

Simplified Pulse-coupled Neural Network for Segmentation of Breast Mass

目前,对乳腺癌的诊断主要靠放射科医生大量读片,容易造成误诊与漏诊。乳腺肿块分割是乳腺癌计算机辅助诊断(CAD)检测和识别系统中关键的一步。本文基于改进型的PCNN神经网络,设置随时间单调上升的阈值,对乳腺肿瘤进行分割,肿块分割完全,分割出的肿块边缘平滑,细节保持较好,能够较好地把目标提取出来。实验结果表明:该方法拥有较好的分割效果。

At present,the diagnosis of breast cancer by radiologists reading a lot of films,it easily lead to misdiagnosis and missed diagnosis.Computer-aided breast mass segmentation is a breast cancer diagnosis（CAD） detection and identification system in a crucial step.This simplified pulse-coupled neural network,set the threshold which increases monotonically time to segment the breast mass,breast mass segmentation complete,separate out the mass edges smooth,detail is maintained well and can better target extracted.Experimental results show that the method has better segmentation effect.