Multimodal Bone Cancer Detection Using Fuzzy Classification and Variational Model Sami Bourouis 1,3 , Ines Chennoufi 2 , and Kamel Hamrouni 1 1 Universit de Tunis El Manar, Ecole Nationale dingnieurs de Tunis 2 ESPRIT : School of Engineering, Tunis, Tunisia 3 Taif University, KSA sami.bourouis@ensi.rnu.tn, ines.channoufi@esprit.tn, kamel.hamrouni@enit.rnu.tn Abstract. Precise segmentation of bone cancer is an important step for several applications. However, the achievement of this task has proven problematic due to lack of contrast and the non homogeneous intensities in many modalities such as MRI and CT-scans. In this paper we inves- tigate this line of research by introducing a new method for segmenting bone cancer. Our segmentation process involves different steps: a regis- tration step of different image modalities, a fuzzy-possibilistic classifica- tion (FPCM) step and a final segmentation step based on a variational model. The registration and the FPCM algorithms are used to locate and to initialize accurately the deformable model that will evolve smoothly to delineate the expected tumor boundaries. Preliminary results show accurate and promising detection of the cancer region. Keywords: Multimodality image fusion, non-rigid registration, fuzzy classification variational model. 1 Introduction Accurate segmentation of bone cancer is an important task for several medical applications. For example, it can be helpful for therapy evaluation, treatment planning, modeling of pathological bones, etc. However, this task is a challenging problem because there is a large class of tumor types which vary greatly in size and position, have a variety of shape and appearance properties, have intensities overlapping with normal bone areas, and may deform and defect the surround- ing structures. Moreover, the majority of images modalities may contain various amounts of noise and artifacts. Traditionally, bone cancers segmentation is per- formed manually by marking the tumor regions by a human expert. This process is time-consuming, impractical and non- reproducible. So, a semi or a fully au- tomatic and robust segmentation is highly required in order to generate quickly satisfactory segmentation results. In general, a single medical image modality cannot provide comprehensive and accurate information, so considering more than one acquisition protocols can provide much more useful information about the bone tumor and this can be achieved through image fusion process. Such pro- cess is used to derive useful information in order to enhance and taking account J. Ruiz-Shulcloper and G. Sanniti di Baja (Eds.): CIARP 2013, Part I, LNCS 8258, pp. 174–181, 2013. c  Springer-Verlag Berlin Heidelberg 2013