基于简化的PCNN在超声乳腺癌图像去噪方面的应用

Applications in the Denoising of Ultrasound Breast Image based on Simplified PCNN

针对乳腺癌超声图像中斑点对诊断的影响,提出一种基于简化的脉冲耦合神经网络(simplified pulse-coupled neuralNetwork,SPCNN)的去噪新方法,并将此方法应用于乳腺癌超声图像滤波。首先利用简化的PCNN定位极端脉冲噪声点并利用中值滤波滤除椒盐噪声,然后利用PCNN赋时矩阵采用分类滤波自适应调节灰度值滤除高斯噪声。用实验图像验证了方法的有效性,然后将此方法应用于乳腺癌的超声图像中进行滤波,实验结果证实该方法对混合噪声在滤波效果和保护细节方面具有优势,对乳腺癌的超声图像能较好地滤除噪声,同时保证了细节,结合医学诊断证实了该方法的有效性。

To put forward a novel denoising method-simplified pulse-coupled neural network(SPCNN),and it was applied in the filtering of breast cancer ultrasound image.First,we used the simplified PCNN to fix the position of extreme impulse noise,salt and pepper noise were removed by the median filter.Second,the Gaussian noise by adjusting the gray value adaptively was removed by using PCNN time matrix with category filter.Experiments show that the method is effective,and it can be applied in the filtering of the ultrasound breast cancer.In addition,the experimental results demonstrate that the method has advantages in the filtering effect and detail preservation for the mixed noise,it is also effective on removing noise and preserving detail information in the breast cancer ultrasound image,it is confirmed that the method has good performance on medical diagnosis.