Tagged cardiac MR image segmentation by contrast enhancement and texture analysis
Xiaomei Yang, Kenya Murase · 2009
Based on contrast enhancement and texture analysis, an automatic segmentation of the left ventricle in 2D tagged MR images is presented in this paper. Incorporating histogram modification and local contrast enhancement is applied as a method for improving contrast between tagged lines and non-tagged tissue. The ventricular blood filled and tagged regions are isolated by subtracting gray minimum from maximum within a small window. The feature value is large for textured region and small for homogeneous region. Media filter is used to effectively remove the influence of the tagged lines. Finally, GVF-snake model is applied to extract the boundary of the left ventricle. Experimental results show that the proposed method can effectively extract the endocardiac boundary of the left ventricle.