A new approach to locate the hippocampus nest in brain MR images

Abstract

Hippocampal shrinkage is a main biomarker for the detection of Alzheimer's disease and Temporal lobe Epilepsy (TLE). Mostly, developing methods for the hippocampus segmentation are unable to initialize automatically due to its low contrast boundary and uncertain position with respect to the wide range of human brain size. This paper will describe how to reduce the search area in brain MRI to determine the hippocampus location by setting a cuboid slice-based nest for the hippocampus called CSNHC surrounding this structure. The proposed algorithm applies a 3D skull stripping method using BET to extract the brain volume, following by the distance estimation from the first slice that brain volume is seen to the first slice including the hippocampus in the coronal, axial and sagittal views. Finally, ground truths for three different dataset including 68 MR images are used to validate our results.