小波多尺度水平集算法与心脏超声图像鲁棒分割

由于斑点噪声的存在,超声图像的灰度分布是非高斯的,传统的基于高斯模型的图像分割方法不能解决心脏超声图像分割问题。但小波分解后的高阶低频小波系数近似服从高斯分布,利用这个特点,论文提出一种新颖的小波多分辨率框架下的水平集曲线演化算法。首先对超声心脏图像做小波分解,得到各层的低频图像。从小波分解的顶层低频图像开始,利用边界和区域复合约束动态轮廓线模型(ActiveContourModel)寻找左心室内边界;然后通过插值将结果向下一尺度低频图像传递,并利用尺度间形状约束和边界约束复合ACM进一步细化曲线,使其符合局部图像特征,如此逐层重复直至原始图像。由于采用了小波多尺度框架和尺度间形状约束,算法具有曲线演化结果稳健鲁棒、不易陷入局部极小和发生边界泄漏等优点,非常适合心脏超声图像噪声高、对比度低、边界灰度梯度不显著的特点。在实际临床三维超声图像上的实验表明,算法分割结果和人工分割结果很接近。

A Novel Wavelet Multi-scale Level Set Algorithm for Robust Segmentation of Echocardiographic Images

Due to the existence of speckle noise in ultrasonic images,the gray level distribution is not Gaussian. Traditional image segmentation methods based on Gaussian model often fail in echocardiographic images.But after wavelet decomposition,the coefficient in high level low frequency sub-band is approximately Gaussian.Based on this characteristic,this paper proposes a novel wavelet multi-scale level set algorithm.Firstly,the echocardiographic image is wavelet transformed to get different scale low frequency approximation images.Then the algorithm begins with the highest level approximation image and outlines the left ventricle endocardium border with regional and edge constrained ACM.Then the result is interpolated into the next finer level of approximation image as a initial contour,and evolved with edge based and inter scales shape constrained ACM.This step is repeated until the finest level is reached.The multi-scale framework and shape constrain make the curve evolution is robust to noise and local minimums,especially effective for noisy and low contrast echocardiographic images with weak edges.Experiments on clinical 3D echocardiographic images show the algorithm’s result is very close to the expert manual outlines.