Reliable extraction of the mid-sagittal plane in 3D brain MRI via hierarchical landmark detection

Reliable detection of the mid-sagittal plane is the key for brain image registration, asymmetry analysis, and group studies. Although the brain presents most of the time a regular structure, outliers in the data consisting of brain tumors or various deformations pose challenges to the existing approaches. We propose in this paper a robust approach for mid-sagittal plane extraction based on hierarchical landmark detection. Cross-validated results demonstrate comparable accuracy (1.08° plane normal error) to those of human experts on a volumetric data set that contains pediatric patients as well as elderly with different diseases.