Automatic Segmentation of Cerebral Hemispheres in MR Human Head Scans P. Kalavathi, 1 V. B. Surya Prasath 2 1 Department of Computer Science and Applications, Gandhigram Rural-Institute Deemed University, Gandhigram 624 302, Tamil Nadu, India 2 Computational Imaging and VisAnalysis (CIVA) Lab, Department of Computer Science, University of Missouri-Columbia, Columbia, MO 65211, USA Received 25 May 2015; revised 28 October 2015; accepted 15 December 2015 ABSTRACT: Automatic segmentation of cerebral hemispheres in magnetic resonance (MR) brain images help to quantify the brain asymmetry and correct several MR brain deformities. The detection of mid-sagittal plane (MSP) in human brain image is necessary to segment the hemispheres for both operator-based and automated brain image asymmetric analysis. In this article, a computationally simple and accurate technique to detect MSP in MRI human head scans using curve fitting is developed. The left and right hemispheres are segmented based on the detected MSP. The accuracy of the MSP is evaluated by comparing the segmented left and right hemi- spheres against the manually segmented ones. Experimental results using 78 volumes of T1, T2 and PD-weighted MRI brain images show that the proposed method has accurately segmented the cerebral hemispheres based on the detected MSP in axial and coronal orien- tations of normal and pathological brain images. V C 2016 Wiley Peri- odicals, Inc. Int J Imaging Syst Technol, 26, 15–23, 2016; Published online in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/ima.22152 Key words: cerebral hemisphere segmentation; mid-sagittal plane; interhemispheric fissure; longitudinal cerebral fissure; MRI brain struc- ture segmentation I. INTRODUCTION The human brain exhibits an approximately bilateral symmetry across the sagittal plane. An inter-hemispheric fissure (IHF) or mid- sagittal plane (MSP) separates the human brain into two distinct cer- ebral hemispheres. However, these two parts are almost never per- fectly symmetric even for the normal brains (Geschwind and Levitsky, 1968; Guillemaud et al., 1995; Toga and Thompson, 2003; Zhao et al., 2009). The automatic detection of MSP in brain mag- netic resonance (MR) images is useful in various medical imaging applications such as volumetric analysis, construction of anatomical models, 3D visualization, multimodal visualization, surgical plan- ning as well as simulations and detection of many brain related dis- eases such as multiple sclerosis, schizophrenia, epilepsy, Parkinson’s disease, Alzheimer’s disease, cerebral atrophy, tumors and other pathologies. Several techniques have been developed for whole brain segmentation (also known as skull stripping) (Somasundram and Kalavathi, 2012a, 2013, 2014; Kalavathi and Prasath, 2015), seg- mentation of various brain structures for automatic diagnosis of brain related diseases such as multiple sclerosis, white matter lesion (Shiee et al., 2010; Ong et al., 2012; Kalavathi, 2013), identification of mid- sagittal plane, and multi-tissue (Moreno et al., 2014) or left right hemispheres segmentations (Liang et al., 2007; Kuijf et al., 2014). There exists wide variety of image segmentation techniques which can be utilized for brain MSP detection (Ma et al., 2009, 2010). Fur- ther, some of the higher order image processing applications of brain images like image registration (Oliveira and Tavares, 2014; Tavares, 2014; Alves and Tavares, 2015) require the detection of MSP as the first step for spatial normalization and/or anatomical standardization of brain images (Minoshima et al., 1994; Lancaster et al., 1995). Automatic detection of MSP simplifies the registration of two- dimensional brain images and bifurcation of left and right cerebral hemispheres (Kapouleas et al., 1991) and is required in many clinical applications such as brain lesions detection (Hojjatoleslami and Kruggel, 2001; Kruggel, 2004). Brain hemispheres segmentation is often needed for various bio- medical and neuro-scientific applications, because most of the brain structures have the bilateral morphology and functional lateralization (Zhao et al., 2010). Moreover, brain asymmetry analysis provides methods for computer-assisted diagnosis for mental diseases such as schizophrenia (Bilder et al., 1994; Petty, 1999; Sommer et al., 2001; Ekin, 2006). For such purposes the location of MSP is ultimately needed to estimate the head orientation. However, the scanning tech- niques do not use any optimal head position as a point of reference and hence a brain slice captured in different orientations may make the process of diagnosis difficult and ambiguous. The presences of various imaging artifacts (Somasundaram and Kalavathi, 2012b) may also make this process more complex. Therefore, an automated Correspondence to: V. B. Surya Prasath; e-mail: prasaths@missouri.edu; surya. iit@gmail.com V C 2016 Wiley Periodicals, Inc.