A Two-Stage Method for Bone X-Rays Abnormality Detection Using MobileNet Network Hadeer El-Saadawy (&) , Manal Tantawi, Howida A. Shedeed, and Mohamed F. Tolba Faculty of Computer and Information Science, Ain Shams University, Cairo, Egypt hadeer_ibrahim@cis.asu.edu.eg Abstract. Abnormality in bones is considered to be from the critical medical cases, since the wrong diagnosis could lead to bad side effects. Moreover, exhausted and over loaded doctors may miss some cases. Hence, Computer aided diagnosis systems have a vital role nowadays. This paper presents a method for detecting the fractures in the seven extremity upper bones (shoulder, humerus, forearm, elbow, wrist, hand and finger) using X-ray images. A two- stage classification method based on MobileNet network is proposed. Enhanced X-ray image is fed into the first stage to detect bone type. Thereafter, the bone image is directed according to the result of the first stage to one of seven classifiers (one for each bone type) to detect the abnormality in the bone. MURA dataset is utilized as a performance dataset and average accuracy 73.42% has been achieved after merging the two classification stages. Keywords: Computer aided diagnosis Á Medical imaging Á Deep learning 1 Introduction Medical imaging techniques have been used in various medical domains. Among the oldest and frequently used domains is the X-Rays domain. Digital X-ray imaging devices are being widely used in many medical applications as a result of the com- bination of radiation and computer image processing technologies [1]. Computer aided bone abnormality detection is utilized to assist the doctors to diagnose the abnormality in human bones. However, bones abnormality detection is a very dif ficult task as the abnormality can occur in any part of human skeleton bones like head bones, trunk bone, upper and lower extremity bones. Hence, the variety in the bones is very large, so the need for computer aided bone abnormality detection for more than one bone type is essential nowadays [2]. X-ray medical imaging is the main tool for the diagnosis of the bone’s abnormality, especially the fractures. X-ray images are utilized by the medical practitioners to be able to take the critical decisions in the diagnosis [2, 3]. Wrong diagnosis of the bone’s abnormality can lead to ineffective patient management, increased dissatisfaction and © Springer Nature Switzerland AG 2020 A.-E. Hassanien et al. (Eds.): AICV 2020, AISC 1153, pp. 372–380, 2020. https://doi.org/10.1007/978-3-030-44289-7_35