Abstract—Image segmentation has a fundamental role in analysis and interpretation for many applications. The automated segmentation of organs and tissues throughout the body using computed imaging has been rapidly increasing. Indeed, it represents one of the most important parts of clinical diagnostic tools. In this paper, we discuss a thorough literature review of recent methods of tumour segmentation from medical images which are briefly explained with the recent contribution of various researchers. This study was followed by comparing these methods in order to define new directions to develop and improve the performance of the segmentation of the tumour area from medical images. Keywords—Features extraction, image segmentation, medical images, tumour detection. I.INTRODUCTION MAGE segmentation is an important technique for segmenting images without overlapping each other and having their own features. It has been rapidly developed in the field of medical imaging. In fact, automated segmentation of organs throughout the body using computed tomography and magnetic resonance imaging has increased rapidly. Among the most widespread applications in the medical field, we can cite the detection of the localization of tumours, estimation of their volume, delineation of the cells, and their surgical planning [1]. Indeed, research in many medical conditions has greatly benefited from these approaches by allowing the development of faster and reproducible quantitative imaging markers. These markers have been used to help diagnose different pathologies, determine the estimate of disease progression, select and monitor appropriate treatments. As some of these tools are moving from research environments to clinical practice, it is important for radiologists to become familiar with the different methods used for automated segmentation. Indeed, this information should help radiologists better evaluate automated segmentation tools and apply them not only to research, but also to clinical practice. This article presents an in-depth study of the literature on recent methods of tumour segmentation from MRI images. It includes performance and quantitative analysis of advanced methods. Different image segmentation methods are briefly Mayssa Bensalah is with the National School of Engineering of Sfax, Tunisia (corresponding author, phone: 021623055131; e-mail: bensalah- maissa@hotmail.fr). Atef Boujelben is with the Faculty of Sciences of Sfax, Tunisia (phone: 0216204183701; e-mail: atef.boujelben@fss.rmu.tn). Mouna Baklouti is with the National School of Engineering of Sfax, Tunisia (phone: 021622819896; e-mail: mouna.baklouti@gmail.com). Mohamed Abid is with the National School of Engineering of Sfax, Tunisia (phone: 021697588722; e-mail: mohamed.abid_ces@yahoo.fr). explained with their advantages and disadvantages, not only for comparative analysis but also for recent contribution from various researchers. This review paper is organized as follows: On the first section, we start by presenting the introduction. The second section is devoted to the presentation of the different methods of segmentation of tumors from medical images. Subsequently, in the third section we elaborate the comparative study with a discussion, and we finally close with a conclusion. II.MEDICAL IMAGES SEGMENTATION Various approaches to automated segmentation of computed tomography (CT) and magnetic resonance (MRI) images are widely used in research environments and promise to transform clinical practice and radiologists involved in image interpretation for patients with cancer, obesity, cardiovascular disease, neurodegeneration, osteoporosis, arthritis, etc. Such approaches will help clinicians diagnose the disease, determine the prognosis, select the patients for treatment and to monitor responses to treatment. To enable this transition from research to patient care, radiologists must become familiar with the different methods used for automated segmentation of CT and MRI images. A. Semiautomatic vs Automatic Segmentation As an indication, segmentation consists in identifying the limits of an object in the image. Frequently, the object is an organ, tissue, pathological lesion or other structure used for the diagnosis or treatment of a particular disease. Traditional approaches to segmentation rely on manual or semi-automated delineation of the object of interest. While these approaches are effective, they are time consuming and impractical for large-scale research studies and even less practical for clinical practice. As a result, many fully automated approaches to tissue segmentation are under development. This is because automated segmentation methods using CT and MRI generally rely on basic image processing of pixel intensities and textural features (relationships between groups of pixels, for example) and may incorporate techniques: advanced model-based, atlas or machine learning. Segmentation techniques can be broadly divided into supervised and unsupervised, as will be highlighted in the next subsection. B. Supervised and Unsupervised Segmentation Supervised segmentation techniques require prior training, usually done manually. These methods typically include pre- processing such as intensity normalization, followed by classification (artificial neural networks, nearest k-neighbours, Mayssa Bensalah, Atef Boujelben, Mouna Baklouti, Mohamed Abid A Comparative Study of Medical Image Segmentation Methods for Tumor Detection I World Academy of Science, Engineering and Technology International Journal of Computer and Information Engineering Vol:15, No:4, 2021 285 International Scholarly and Scientific Research & Innovation 15(4) 2021 ISNI:0000000091950263 Open Science Index, Computer and Information Engineering Vol:15, No:4, 2021 publications.waset.org/10011999/pdf