基于深度学习的人工智能在肺结节检测领域的研究进展

Research Progress of Deep Learning-Based Artificial Intelligence in Pulmonary Nodules Detection

随着人民生活水平提高、居民保健意识增强和低剂量CT在健康体检的广泛应用,肺结节的检出率逐渐增加。胸部薄层CT检查产生大量的影像数据,对胸部影像医师日常工作提出严峻挑战。人工智能(artificial intelligence,AI)在影像诊断中的应用,能快速发现肺内结节,一定程度减轻了影像医生的繁重工作。不同的AI模型在检出和诊断肺结节的敏感性和准确性方面需要在临床应用中逐渐完善提高。深度学习是其中一种较为重要的方法,通过收集海量的数据并建立独特的模型,进而反复验证模型的准确性,最终获得高效率的检测系统。近年来,深度学习领域引起了人们的兴趣,基于深度学习的医学成像领域的最新研究成果令人振奋。本文对目前深度学习在肺结节病变中的研究进展进行详细描述。

With the improvement of people’s living standard,the enhancement of residents’health awareness and the wide application of low-dose CT in health examination,the detection rate of pulmonary nodules has gradually increased.Chest thin slice CT examination produces a large amount of image data,which poses a severe challenge to the daily work of chest imaging physicians.The application of artificial intelligence(AI)in imaging diagnosis can quickly find pulmonary nodules,and relieve the heavy work of imaging doctors to a certain extent.The sensitivity and accuracy of different AI models in the detection and diagnosis of pulmonary nodules need to be gradually improved in clinical application.Deep learning is one of the more important methods.By collecting massive data and establishing unique models,and then repeatedly verifying the accuracy of the models,an efficient detection system is finally obtained.In recent years,the field of deep learning has aroused people’s interest,and the latest research results in the field of medical imaging based on deep learning are exciting.In this paper,the current research progress of deep learning in pulmonary nodules is described in detail.