TECHNICAL PAPER Automatic MR image segmentation using maximization of mutual information Apurba Roy 1 • Santi P. Maity 2 Received: 22 July 2017 / Accepted: 5 July 2018 Ó Springer-Verlag GmbH Germany, part of Springer Nature 2018 Abstract Magnetic resonance (MR) brain image segmentation is an important task for the early detection of any deformation followed by the quantitative analysis for the prediction and stage defection of brain diseases. But segmentation of the MR brain image suffers from limited accuracy as captured images have non-uniform homogeneity over an organ, presence of noise, uneven and broken boundary etc. Due to the complex structure of the brain and varieties of the captured MR images, only a single feature based MR image segmentation cannot give sufficient accurate result. In the proposed method thresholds for segmenting the MR image are computed by maximizing the mutual information for the two features, compactness and homogeneity. The proposed algorithm is tested against the real T1 MR image to asses the accuracy. Further the output is validated and compared with the ground truth and other recently reported works. 1 Introduction Segmentation of the MR image is an important and chal- lenging problem for the medical image analysis to offer an accurate diagnosis and consequent treatment planning. Alzheimers Mathew et al. (2016) and schizophrenia Song et al. (2016) are the two common examples of the brain disease. Alzheimers disease indicates a state where neurons within the brain do not work properly, lose the connectivity within them and ultimately those neurons die due to lack of food and oxygen. Alzheimer’s disease is the most common cause of dementia where the patients’ brains do not work properly for different functions, thinking, memory power, judgment etc. Schizophrenia is a psychotic illness where patients suffer from difficulty to organize thoughts and are not able to show emotional expression. Schizophrenia and alzheimers disease are both associated with psychotic symptoms of the brain that cause hallucinations or delu- sions. In hallucinations, patients suffer from false sensory perceptions like they smell, touch, hear and view things which are not real. On the other hand, patients suffering from delusions have some fixed false believes that are not happening in reality. Over the last one decade there have been rapid progress in computer aided diagnosis on med- ical images for proper treatment of the diseases caused in the human body including brain, heart etc. Different imaging modalities are developed that include computer tomography scan, magnetic resonance imaging, positron emission tomography imaging, X-Ray etc. Each imaging scheme offers some advantages and disadvantages too. Furthermore, a particular imaging technology is found to be useful for capturing images of certain type. MR imaging has become very useful as it is invasive in nature and also shows good response against soft tissues. Though MR images are found to be efficient due to their soft tissue analysis and invasive in nature, their segmentation become difficult and challenging due to uneven and broken boundary creating complex tissue structure, non-uniform homogeneity over a tissue etc. To that aim, this work aims to develop an automatic MR image segmentation scheme. 1.1 Literature review and scope of the work There are extensive research works reported in the litera- ture that deal with the segmentation of the MR images & Apurba Roy apurbaroy@cemk.ac.in Santi P. Maity santipmaity@it.iiests.ac.in 1 Department of Information Technology, College of Engineering and Management, Kolaghat, West Bengal 721171, India 2 Department of Information Technologies, Indian Institute of Engineering Science and Technology, Shibpur, West Bengal 711103, India 123 Microsystem Technologies https://doi.org/10.1007/s00542-018-4031-y