Segmentation for Lenke Scoliosis X-Ray Images Using Machine Learning

Lenke is a classification of spinal deformities first discovered by Dr. Lwrence G. Lenke, and since then, new classification systems have been used to assess the severity of scoliosis. According to Lenke's classification criteria, scoliosis is divided into 6 types, namely type 1 to type 6. In reading the severity of scoliosis, doctors make observations through X-Ray images. This method takes time and is tiring to find out in detail the type and diagnose existing disorders. This can cause observation results to be inaccurate and allow misdiagnosis to occur. Currently, several studies have been carried out to develop machine learning technology for Lenke scoliosis classification based on machine learning. The stage that determines the ease of classification is segmentation to separate objects from their background, making it easier for doctors to determine existing abnormalities. Segmentation in Lenke scoliosis types 1–5 is still a great opportunity for development. In this research we will segment X-Ray images of Lenke scoliosis types 1–5 using machine learning, namely Thresholding and Active Contour. By comparing the visual results and the Dice Coefficient, Precision, Recall and Standard Deviation metrics, the research shows that the segmentation method active contour is slightly better than the threshold method.

I. Introduction

Lenke scoliosis is a spinal disorder first proposed by Dr. Lwrence G. Lenke, known as the “Lenke Classification System”. In research from Lenke, et al.[1], proposed a new classification system to assess the severity of scoliosis, known as as the Lenke classification criteria, and these have since become standard guidelines for evaluating scoliosis in clinical practice, the Lenke classification criteria divide scoliosis into six types, from Lenke type 1 to type 6. Obtaining an accurate Lenke classification of scoliosis is essential for selecting treatment modalities, especially operational strategy. Scoliosis itself is a condition in which the spine forms a curve that exceeds 10 degrees, [2] and is visible via direct posterioanterior radiography. According to Woods C. G., the curvature causes deformity not only in the coronal plane, but also involves all three planes, triggered by the self-rotational movement of the spine. Approximately 80% of structural coronal deformities are referred to as idiopathic scoliosis [3]. says The process of diagnosing idiopathic scoliosis involves excluding known causes. Idiopathic scoliosis is divided into three subgroups based on age, namely infantile (ages 0–3), juveniles (ages 4–9), and adolescents (ages 10 to adulthood). [4] [5]. X-ray examination of the entire spine is the most commonly used imaging examination to diagnose, treat, and provide a prognosis in cases of scoliosis. Evaluation of the Cobb angle (a type of measurement of the lateral curvature of the spine), vertebral rotation, and other parameters in X-rays can effectively reflect the severity of scoliosis and provide a basis for formulating the best treatment plan [6].