论文部分内容阅读
机器学习作为人工智能的主要技术方向，可帮助眼科医生解读与分析成像设备产生的大量数据，简化诊疗过程。圆锥角膜的分类和早期诊断是机器学习的一个重要应用实例。机器学习用于辅助诊断圆锥角膜的建模方式通常有神经网络法、决策树法，这些模型的敏感性和特异性均在85%以上，但由于圆锥角膜的研究参数不一致，且缺乏公共数据集来衡量算法的优劣，限制了其在临床上的普遍推广。角膜屈光手术术前评估存在数据量大、决策困难的临床问题，机器学习可辅助评估患者是否适合进行屈光手术，其特异性、敏感性均在90%以上，并可通过术前各种眼部参数预测术后视觉质量。另外机器学习在角膜内皮细胞密度计数、角膜上皮损伤程度评估方面都有应用。通过机器学习及大数据建模可协助医生进行角膜病的精准诊断和个性化评估，为角膜病诊疗奠定数据基础。本文对近年来机器学习在角膜相关疾病中的应用进展进行综述。“,”Machine learning, as the main technical direction of artificial intelligence, can help ophthalmologists to interpret and analyze the large amount of data generated by imaging equipment, and also simplify the diagnosis and treatment process.The early diagnosis and classification of keratoconus became the most important application of machine learning.The modeling methods of machine learning for diagnosing keratoconus usually included neural network and decision tree method.The sensitivity and specificity of these models for diagnosing keratoconus were more than 85%.Because there were large number of research parameters for the diagnosis of keratoconus and no adequate public data sets, it was difficult to evaluate the advantages and disadvantages of different research methods, which limited the clinical application of machine learning in the evaluation of keratoconus.The corneal refractive surgery preoperative evaluation had clinical problems of large data volume and difficult decision-making.Machine learning can assist in evaluating whether the patient is suitable for refractive surgery, of which specificity and sensitivity were above 90%.It was also able to predict postoperative visual quality with ocular parameters.In addition, machine learning can also help us to count corneal endothelial cell density and assess corneal epithelial damage.Machine learning method and big data modeling evaluation can assist doctors in accurate diagnosis and personalized evaluation of keratopathy.This article reviewed the recent literature on the application progress of machine learning in corneal-related diseases in recent years.