Local region statistics combining multi-parameter intensity fitting module for medical image segmentation with intensity inhomogeneity and complex composition

• Proposes a novel local region-based model for medical image segmentation.
• Uses the local region statistics and multi-parameter intensity fitting to construct the energy function.
• Segmentation and bias field estimation can be jointly obtained.

It is difficult to segment medical image with intensity inhomogeneity and complex composition, because most region-based modules relay on the intensity distributions. In this paper, we propose a novel method which uses local region statistics and multi-parameter intensity fitting as well. By replacing the original local region statistics with the novel local region statistics after bias field correction, the effect of intensity inhomogeneity can be eliminated. Then we devise a maximum likelihood energy function based on the distribution of each local region. Segmentation and bias field estimation can be jointly obtained by minimizing the proposed energy function. Furthermore, in order to characterize the features of each local region effectively, two parameters are used to fit the average intensity inside and outside of the counter, respectively. This can well handle the medical images with complex composition, such as larger gray difference even in the same region. Comparisons with several representative methods on synthetic and medical images demonstrate the superiority of the proposed method over other representative algorithms.