Zhou Bo, Ding Victoria Y, Li Ying, Ball Robyn L, Jiang Bin, Zhu Guangming, Boothroyd Derek, Zeineh Michael, Gean Alisa, Wintermark Max
From the Department of Radiology, Neuroradiology Section, Stanford University, Stanford, CA.
Department of Neurology, the Second Medical Centre, National Clinical Research Center for Geriatric Diseases, Chinese PLA General Hospital, Beijing, China.
J Comput Assist Tomogr. 2019 Sep/Oct;43(5):690-696. doi: 10.1097/RCT.0000000000000913.
The aim of the study was to refine and validate the NeuroImaging Radiological Interpretation System (NIRIS), which was developed to predict management and clinical outcome based on noncontrast head computerized tomography findings in patients suspected of acute traumatic brain injury (TBI).
We assessed the performance of the NIRIS score in a prospective, single-center cohort of patients suspected of TBI (n = 648) and compared the performance of NIRIS with that of the Marshall and Rotterdam scoring systems. We also revised components of the NIRIS scoring system using decision tree methodologies implemented on pooled data from the retrospective and prospective studies (N = 1190).
The NIRIS performed similarly to the Marshall and Rotterdam scoring systems in predicting mortality and markedly better in terms of predicting more granular elements of disposition and management of TBI patients, such as admission, follow-up imaging, intensive care unit stay, and neurosurgical procedures. The revised NIRIS classification correctly predicted disposition and outcome in 91.2% (331/363) after excluding patients with other major extracranial traumatic injuries or intracranial nontraumatic injuries.
The present study further demonstrates the predictive value of NIRIS in guiding standardized clinical management and decision-making regarding treatment options for TBI patients.
本研究旨在完善并验证神经影像放射学解释系统（NIRIS），该系统旨在根据疑似急性创伤性脑损伤（TBI）患者的非增强头部计算机断层扫描结果预测治疗方案和临床结局。
我们在一个前瞻性、单中心的疑似TBI患者队列（n = 648）中评估了NIRIS评分的表现，并将NIRIS的表现与马歇尔评分系统和鹿特丹评分系统进行了比较。我们还使用决策树方法对NIRIS评分系统的组成部分进行了修订，该方法应用于回顾性和前瞻性研究的汇总数据（N = 1190）。
在预测死亡率方面，NIRIS的表现与马歇尔评分系统和鹿特丹评分系统相似，而在预测TBI患者更具体的处置和管理要素方面，如入院、随访成像、重症监护病房住院时间和神经外科手术等，NIRIS的表现明显更好。在排除患有其他主要颅外创伤性损伤或颅内非创伤性损伤的患者后，修订后的NIRIS分类正确预测处置和结局的比例为91.2%（331/363）。
本研究进一步证明了NIRIS在指导TBI患者治疗方案的标准化临床管理和决策方面的预测价值。
