结合人工智能诊断青少年特发性脊柱侧弯的新思路

Combining artificial intelligence for diagnosing adolescent idiopathic scoliosis

青少年特发性脊柱侧弯(adolescent idiopathic scoliosis,AIS)是脊柱侧弯的最常见类型,多发生在青春期,表现为脊柱向侧方弯曲>10°。如能早期诊断、及时干预,可显著减少并发症及改善预后。对于AIS的校园筛查虽然备受争议,但亦被广泛推行。目前评估脊柱侧弯的最基本方法是使用脊柱侧弯测量尺在Adam前屈试验中进行临床检查。金标准仍然是1948年被提出的基于X线平片人工测量Cobb角,但该诊断方式具有辐射性,且十分依赖检查者的专业能力,因此寻找灵敏度高、特异度高、无辐射、高效的早期筛查及诊断方法尤为迫切。本文围绕基于超声系统的三维脊柱成像、基于三维相机或传感器的脊柱侧弯检测系统、平面数码相机结合人工智能、基于计算机视觉的姿势分析系统等方面对人工智能应用于AIS筛查、诊断的研究进展进行综述,为相关临床科学研究及发展方向提供新思路。

As the most common type of scoliosis,adolescent idiopathic scoliosis(AIS)occurs predominantly in adolescents.Lateral spinal curvature>10°is the most common type of scoliosis.An early diagnosis and a timely intervention are essential for reducing complications and optimizing outcomes.Although mass screening for AIS has remained controversial at schools,it is also widely implemented.Currently a routine tool for evaluating scoliosis is using a scoliometer for Adam test.First proposed in 1948,manual radiographic measurement of Cobb angle has remained a gold standard.However,this diagnostic modality is radioactive and it relies heavily on the professional capability of an examiner.Thus it is quite imperative to seek early screening and diagnostic modalities with high sensitivity,high specificity,non-radiation and high efficiency.In recent years,with a rapid development of artificial intelligence(AI),it has also accelerated the progress of medical imaging and diagnostics.This review focused upon three-dimensional spinal imaging based upon ultrasound,scoliosis detection system based upon three-dimensional camera or sensor,combination of two-dimensional digital camera and AI and posture analysis system based upon computer vision for applying artificial intelligence to AIS screening.It was intended to offer new rationales for clinical scientific researches and development directions.