摘要
中国有世界上第二大的结核病疫情(仅次于印度),具有较高的结核病感染、发病、耐药以及病死率。因此,快速、准确的诊断,以及时采取治疗,是控制这种疾病的关键问题。涂片荧光显微镜与金胺-O染色是在病理实验室检测肺结核抗酸杆菌(acid fast bacilli,AFB)最常用的诊断方法。但是以眼睛观察显微镜来筛查检测抗酸杆菌是一项烦琐、劳动密集型任务。低质量、不一致的痰涂片染色技术,标本碎屑,人眼视觉的变异和疲劳等因素会导致灵敏度低至40%。使用基于人工智能(artificial intelligence,AI)计算机辅助诊断(computer aided diagnostic,CAD)技术的自动显微镜系统对结核病进行自动诊断,提供了一个有效解决当前痰涂片诊断结核病所存在的缺陷。胸部X射线片也是世界卫生组织认定的非常有效的快速分流和转诊检测方法。但是,在不发达地区,因放射医生缺乏,它无法为大量的感染人群提供服务。为解决这一需要,将先进的数字化医学影像精准诊断应用于肺结核(pulmonary tuberculosis,PTB)检测,基于人工智能的CAD自动化智能系统,为肺结核的数字化医学影像精准诊断开辟了一条新路。
China has the world's second largest tuberculosis epidemic(after India)with very high TB infection,TB incident,drug resistance,and mortality rate.Rapid,accurate diagnosis is critical for timely initiation of treatment and ultimately control of the disease.WHO-recommended smear fluorescent microscopy is the most common diagnostic tools in the laboratories to detect acid fast bacilli(AFB)after staining with Auramine-O.Routine visual slide screening for identification and counting of AFB is a tedious,labor-intensive task.Low quality,inconsistent slide staining technique,debris,variation in human perception,and fatigue lead to sensitivity as low as 40%,especially in scanty specimens.Applying an automatic microscopy system using artificial intelligence based computer aided diagnostic(CAD)technologies to the automated diagnosis of TB presents the opportunity to address the shortcomings of current techniques in diagnosing TB from sputum smears.For the identification of TB suspects in low-resource settings,WHO has recommended chest X-ray(CXR)screening as a very efficient triage referral test.A challenge in those regions,however,is the imbalance in the affected population and available radiology services.In addressing this need,application of CAD using artificial intelligence into a low-cost automated tool for pulmonary TB(PTB)in CXR images can directly close this gap.
出处
《新发传染病电子杂志》
2017年第1期5-9,共5页
Electronic Journal of Emerging Infectious Diseases
基金
Multi-Modality Image Data Fusion and Machine Learning Approaches for Personalized Diagnostics and Prognostics of MCI due to AD.1R41AG053149-01A1(10/2016-8/2017)Agency:HHS/NIH/NoA
A dual-polarized Doppler Radar system for fall detection in an indoor environment。1R41AG050382-01(9/2015-8/2017)Agency:HHS/NIH/NoA
关键词
肺结核
自动化显微镜
数字影像学
计算机辅助诊断
人工智能
Pulmonary tuberculosis
Automated microscopy
Digital radiography
Computer aideddiagnosis
Artificial intelligence