S.I. : EMERGING INTELLIGENT ALGORITHMS FOR EDGE-OF-THINGS COMPUTING Computer-based Cobb angle measurement using deflection points in adolescence idiopathic scoliosis from radiographic images Areen K. Al-Bashir 1 • Mohammad A. Al-Abed 2 • Hala K. Amari 1 • Fadi M. Al-Rousan 3 • Omar M. K. Bashmaf 3 • Enas W. Abdulhay 1 • Rabah M. Al Abdi 1 • N. Arunkumar 4 • B. R. Tapas Bapu 5 • Ahmad K. Al-Basheer 6 Received: 23 April 2018 / Accepted: 25 June 2018 Ó The Natural Computing Applications Forum 2018 Abstract Idiopathic scoliosis treatment depends on the accurate assessment of the Cobb angle, which is usually performed manually. Manual measurements, however, can lead to observer variations, which depend on the correct selection of the curvature superior and inferior vertebrae in order to draw the needed lines for Cobb angle measurements. In this paper, we are proposing an algorithm to measure the Cobb angle semi-automatically. The algorithm is based on two processing phases in which each column in the raw X-ray image is reduced to two points representing the end points of the spine and containing its general structure and outline. These points are then used to fit a fifth-order polynomial. We hypothesize that the deflection points of the fitted curve represent the superior and inferior vertebrae of the scoliosis curvature. The deflection points were used to calculate the Cobb angle. The algorithm was tested on X-ray images from 28 subjects (14 females and 14 males, average age of 15.6 ± 1.3 years) diagnosed with adolescence idiopathic scoliosis. Three manual measurements were obtained, with manually measured Cobb angles ranging from of 10° to 98°. The mean of the standard deviation of the manual readings and the algorithm results was 5.28° and 2.64°, respectively, with mean abs error of 6.6° and R value of 0.81. Excluding the cervical and rib cage touching scoliosis cases, the mean of the standard deviation of the manual readings and the algorithm results was 4.73° and 2.35°, respectively, with mean abs error of 3.78° and R value of 0.94. From the results, we can conclude that our proposed algorithm can minimize and simplify user intervention, thus allowing easier and more accurate Cobb angles measurements and resulting in a shorter diagnosis time and requiring no special skills from the user. Keywords Adolescence idiopathic scoliosis Á Digital radiography Á Deflection points Á Cobb angle Á Automation 1 Introduction Idiopathic scoliosis (IS) is defined as three-dimensional torsional deformity of the spine column, generally char- acterized by a lateral deviation of the spine, combined with variable degrees of rotational as well as translational deformity of the spine [1–5]. Diagnosis of the deformity is usually done by a standing radiograph of the spine, which is regarded as the gold standard for the diagnosis and monitoring of such cases. From these radiographs, a Cobb’s angle can be obtained [1, 6, 7], which is defined as the angle formed between a line drawn through the upper part of superior vertebrae and a second line drawn through the lower part of inferior vertebrae, and this angle is used to evaluate the degree of lateral curvature of the spine [1, 6, 8]. A Cobb angle larger than 10° is considered clinically significant for the diagnosis of IS [1, 5]. & Areen K. Al-Bashir akbashir@just.edu.jo 1 Biomedical Engineering Department, Jordan University of Science and Technology, P. O. Box 3030, Irbid 22110, Jordan 2 Biomedical Engineering Department, the Hashemite University, Zarqa 13133, Jordan 3 Orthopedic Department, Jordanian Royal Medical Services, Amman 11855, Jordan 4 Department of Electronics and Instrumentation, Sastra University, Thajavur, India 5 Faculty of Electronics and Communication Engineering, S.A. Engineering College, Chennai, India 6 Radiation Oncology Department, Medical College of Georgia, Augusta, GA 30912, USA 123 Neural Computing and Applications https://doi.org/10.1007/s00521-018-3614-y