PROJECT TITLE :

Nonparametric joint shape and feature priors for segmentation of Dendritic spines - 2016

ABSTRACT:

Multimodal form density estimation is a difficult task in many biomedical image segmentation problems. Existing techniques within the literature estimate the underlying shape distribution by extending Parzen density estimator to the house of shapes. Such density estimates are solely expressed in terms of distances between shapes that could not be sufficient for guaranteeing correct segmentation when the observed intensities offer terribly little information regarding the object boundaries. In such situations, using additional shape-dependent discriminative features as priors and exploiting each shape and have priors can aid to the segmentation method. During this paper, we propose a segmentation algorithm that uses nonparametric joint form and have priors using Parzen density estimator. The joint previous density estimate is expressed in terms of distances between shapes and distances between options. We incorporate the learned joint shape and have previous distribution into a maximum a posteriori estimation framework for segmentation. The resulting optimization downside is solved using active contours. We have a tendency to gift experimental results on dendritic spine segmentation in two-photon microscopy images that involve a multimodal shape density.