基于NSST与改进模糊的乳腺超声图像增强方法  

Breast Ultrasound Image Enhancement Method Based on Combination of NSST and Improved Fuzzy

作　　者：陈雅玲 童莹[2] 何睿清 曹雪虹[2] CHEN Yaling;TONG Ying;HE Ruiqing;CAO Xuehong(School of Telecommunications and Information Engineering,Nanjing University of Posts and Telecommunications,Nanjing Jiangsu 210003,China;School of Information and Communication Engineering,Nanjing Institute of Technology,Nanjing Jiangsu 211167,China)

机构地区：[1]南京邮电大学通信与信息工程学院,江苏南京210003 [2]南京工程学院信息与通信工程学院,江苏南京211167

出　　处：《中国医疗设备》2023年第7期28-33,49,共7页China Medical Devices

基　　金：国家自然科学基金(61703201;61905108);江苏省高等学校自然科学研究面上项目(19KJB140010);南京工程学院科研基金(ZKJ202003);南京工程学院引进人才科研启动基金项目(YKJ201868);南京市科学技术局生命健康科技专项(202110030)。

摘　　要：目的 为解决乳腺超声图像在采集和传输过程中引入噪声导致图像质量下降,影响乳腺癌早期诊断的问题,提出一种基于非下采样剪切波变换(Non-Subsampled Shearlet Transform,NSST)与改进模糊的乳腺超声图像增强方法。方法 首先,通过改进模糊算法增强图像对比度;然后,采用NSST将图像分解为低频部分和高频部分,其中对低频部分进行线性变换以调整图像整体对比度,对高频部分采用阈值模型去除图像中的噪声;最后,将处理后的高频部分和低频部分通过逆NSST获得增强图像。采用信噪比(Signal to Noise Ratio,SNR)和对比噪声比(Contrast to Noise Ratio,CNR)衡量算法去噪性能,结构相似性、特征相似性和信息熵衡量算法细节保留能力,平均梯度衡量算法对比度增强效果。结果 本文方法增强后图像的SNR为2.108,CNR为0.903,信息熵为7.363,平均梯度为9.439,结构相似性为0.939,特征相似性为0.972,均明显高于基于非局部均值自适应选择搜索区域图像去噪算法、基于NSST与模糊对比度的增强算法和基于双边滤波的NSST去噪算法。结论 该图像增强方法在保证图像结构的条件下,能有效降低噪声,增强图像对比度,具有一定的实际应用价值。Objective To solve the problem that noise introduced in the acquisition and transmission of breast ultrasound images may decrease the image quality and affect the early diagnosis of breast cancer,to propose a breast ultrasound image enhancement method based on non-subsampled shearlet transform(NSST)and improved blur.Methods Firstly,the image contrast was enhanced by improved fuzzy algorithm.Secondly,the image was decomposed into low-frequency and high-frequency parts by inverse NSST transformation,the low-frequency part was linearly transformed to adjust the overall contrast of the image,and the threshold model was used to remove the noise in the high frequency part.Finally,the processed high and low frequency parts were inversely transformed by NSST to obtain an enhanced image.Using signal to noise ratio(SNR)and contrast to noise ratio(CNR)to measure the denoising performance of the algorithm,structural similarity,feature similarity and information entropy to measure the algorithm’s detail retention ability,average gradient to measure the algorithm’s contrast enhancement effect.Results By using the method presented in this paper,the SNR of enhanced image was 2.108,the CNR was 0.903,the information entropy was 7.363,the average gradient was 9.439,the structural similarity was 0.939,and the feature similarity was 0.972,all of which were significantly higher than the adaptive selection of search region for NLM based image denoising,the enhancement algorithm based on NSST and fuzzy contrast,and the NSST denoising algorithm based on bilateral filtering.Conclusion The image enhancement method can effectively reduce noise and enhance image contrast under the condition of ensuring the image structure.The research results has practical value.

关 键 词：图像增强 乳腺超声图像 改进模糊算法 非下采样剪切波变换 噪声 

分 类 号：R445.1[医药卫生—影像医学与核医学] TP391[医药卫生—诊断学]