Variational Level Set Segmentation and Bias Correction of Fused Medical Images

Medical image fusion and segmentation has high impact on the digital image processing due to its spatial resolution enhancement and image sharpening. It has been used to derive useful information from the medical image data that provides the most accurate and robust method for diagnosis. This process is a compelling challenge due to the presence of inhomogeneities in the intensity of images. For addressing this challenge, the region based level set method is used for segmenting the fused medical images with intensity inhomogeneity. First, the IHS-PCA based fusion method is employed to fuse the images with intensity inhomogeneity which is then filtered using the homomorphic filter. Then based on the model of the fused image and the derived local intensity clustering property, the level set energy function is defined. This function is minimized to simultaneously partition the image domain and to estimate the bias field for the intensity inhomogeneity correction. The outstanding performance of this approach is illustrated using images of various modalities. Experimental results highlight the effectiveness and advantage of this approach with the help of various metrics and the results are found to be good and accurate.