Latha S, Samiappan Dhanalakshmi
Electronics and Communication Engineering Department, SRM Institute of Science and Technology, Kancheepuram, Tamil Nadu, India.
Curr Med Imaging Rev. 2019;15(4):414-426. doi: 10.2174/1573405614666180402124438.
Carotid artery images indicate any presence of plaque content, which may lead to atherosclerosis and stroke. Early identification of the disease is possible by taking B-mode ultrasound images in the carotid artery. Speckle is the inherent noise content in the ultrasound images, which essentially needs to be minimized.
The objective of the proposed method is to convert the multiplicative speckle noise into additive, after which the frequency transformations can be applied.
The method uses simple differentiation and integral calculus and is named variable gradient summation. It differs from the conventional homomorphic filter, by preserving the edge features to a great extent and better denoising. The additive image is subjected to wavelet decomposition and further speckle filtering with three different filters Non Local Means (NLM), Vectorial Total Variation (VTV) and Block Matching and 3D filtering (BM3D) algorithms. By this approach, the components dependent on the image are identified and the unwanted noise content existing in the high frequency portion of the image is removed.
RESULTS & CONCLUSION: Experiments conducted on a set of 300 B-mode ultrasound carotid artery images and the simulation results prove that the proposed method of denoising gives enhanced results as compared to the conventional process in terms of the performance evaluation methods like peak signal to noise ratio, mean square error, mean absolute error, root mean square error, structural similarity, quality factor, correlation and image enhancement factor.
颈动脉图像显示斑块内容的存在，这可能导致动脉粥样硬化和中风。通过采集颈动脉的B超图像可以实现对该疾病的早期识别。斑点是超声图像中固有的噪声内容，本质上需要将其最小化。
所提方法的目的是将乘性斑点噪声转换为加性噪声，之后即可应用频率变换。
该方法使用简单的微分和积分运算，名为可变梯度求和。它与传统的同态滤波器不同，在很大程度上保留了边缘特征且去噪效果更好。对加性图像进行小波分解，并使用三种不同的滤波器——非局部均值（NLM）、矢量全变差（VTV）和块匹配与三维滤波（BM3D）算法进行进一步的斑点滤波。通过这种方法，识别出依赖于图像的成分，并去除图像高频部分中存在的不需要的噪声内容。
对一组300幅B超颈动脉图像进行的实验以及模拟结果证明，在所提去噪方法与传统方法的对比中，就峰值信噪比、均方误差、平均绝对误差、均方根误差、结构相似性、品质因数、相关性和图像增强因子等性能评估方法而言，所提方法能给出更好的结果。
