Detection the mid-sagittal plane in brain slice MR images by using local ternary pattern

The detection and localization of the mid-sagittal plane (MSP) in brain MR images is an important step in the diagnosis of brain pathologies and calculation of the symmetry between the two sides of a MSP. In this paper, a new method is proposed to locate MSP in T1-weighted MRI images, based on the computation of the different local texture between inter-hemispheric fissure (IF) region and the surrounding tissue. This method is based on local binary pattern technique. It is high-performance texture features to describe local structures. However, the LBP method is sensitive to noise. To overcome this disadvantage, we proposed a new method by using Block Local Ternary Patterns for achieving high classification accuracy thus improving the encoding of the texture feature. The proposed method is based on the second-moments method to solve the tilted brain problem; the rotation angle resulted is used to determine the initial line of the MSP. The performances of the proposed method are compared with the two extensions of LBP features, average LBP and block-based LBP. As it can be seen from the experimental data, the efficiency of the proposed method is better in comparison to the traditional LBP techniques in terms of classification accuracy.