基于OMP和K-SVD的右心室图像分割算法研究

Research on right ventricular image segmentation algorithm based on OMP and K-SVD

为提高对右心室形态结构和功能异常的检测和诊断对心脑血管疾病的作用,并针对其心腔几何形状复杂,解剖结构特殊,使心脏磁共振图像准确分割变成难点,提出通过OMP和K-SVD的有效结合方法。首先将字典学习过程和稀疏表示结合,这样能更好的训练结果,然后由这些训练出的字典模型和稀疏表示系数进行图像的特征重构,最后根据重构后的误差进行分类实现分割。以MICCAI右心室分割挑战中提供的数据为依托,按照如上方法对右心室进行分割,并对分割结果进行分析,实验结果证明提出的基于K-SVD训练字典的分割方法,分割结果较为准确,从而表明了此方法的有效性。

To improve the structure and function of right ventricular morphology of anomaly detection and diagnosis of cardiovascular and cerebrovascular diseases,and aims at its heart chamber geometry is complex,special anatomical structures,and an exact cardiac magnetic resonance image segmentation into difficulties,put forward by combining the OMP and K-SVD method.Dictionary of the learning process is combined with sparse representation,it can better training result,then the dictionary by these training model and the characteristics of the coefficient of sparse representation for image reconstruction,the division of classified according to the error of reconstruction after the implementation.With MICCAI right ventricle segmentation based on data provided by the challenge,in accordance with the above method to split,the right ventricle and the segmentation result is analyzed,the experimental results show that the proposed segmentation method based on K-SVD training dictionary,segmentation result is relatively accurate,so as to show the effectiveness of this method.