相关实验视频

Updated: May 6, 2026

Stereotactic Atlas-Guided Laser Capture Microdissection of Brain Regions Affected by Traumatic Injury

Stereotactic Atlas-Guided Laser Capture Microdissection of Brain Regions Affected by Traumatic Injury

Published on: September 11, 2017

在创伤性脑损伤放射学报告中,利用语言模型进行自动标签生成

Lingrui Cai¹, Craig Williamson², Andrew Nguyen²

¹Gilbert S. Omenn Department of Computational Medicine and Bioinformatics, University of Michigan, Ann Arbor, MI, USA.

Research square

|December 11, 2025

概括

自动化自然语言处理 (NLP) 框架从头部CT扫描中提取关键发现,改善创伤性脑损伤 (TBI) 诊断. 这项技术可以更快,更一致地解释放射学报告,以获得更好的神经创伤护理.

科学领域:

医疗信息学医疗信息学
人工智能在医学中的应用
放射学放射学是一门学科.

背景情况:

及时解释头部CT扫描对于管理创伤性脑损伤 (TBI) 至关重要.
放射学报告的延迟可能会阻碍紧急的临床决策.
目前分析放射学报告的方法通常是手动的,耗时的.

研究的目的:

开发和评估自然语言处理 (NLP) 框架,自动将自由文本头部CT放射学报告转换为结构化,机器可读的结果.
为了提高临床发现的准确性和效率,并从放射学报告中提取位置.
为了能够更快,更一致地解释放射学报告,以改善神经创伤护理.

主要方法:

利用了4038份未识别的头部CT报告,以及444份专家注释的样本.
对比了各种NLP策略,包括词汇加权域适应性预训练和位置感知级联模型.
雇员半监督学习使用未标记的报告来评估绩效增长.

主要成果:

一种词典加权的域适应性预训练方法在五倍交叉验证中获得了0.92的加权F1得分.
位置感知级联模型增强了解剖部位的识别,提高了透明度和临床相关性.
半监督学习比标准监督模型提供了适度的绩效改进.

结论:

在NLP中,特定领域的适应和结构化建模可以从放射学文本中可靠地提取关键发现.
这些NLP框架有助于更快,更一致地解释放射学报告.
自动报告生成可以缩短报告处理时间,并支持数据驱动的神经创伤护理.

关键词:

临床决策支持临床决策支持域名适应域名适应自然语言处理自然语言处理放射学报告放射学报告半监督学习半监督学习创伤性脑损伤创伤性脑损伤

更多相关视频

Semi-quantitative Assessment Using [18F]FDG Tracer in Patients with Severe Brain Injury

Semi-quantitative Assessment Using [18F]FDG Tracer in Patients with Severe Brain Injury

Published on: November 9, 2018

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Published on: December 6, 2024

相关概念视频

Traumatic Brain Injury l: Introduction

Traumatic Brain Injury l: Introduction

DefinitionTraumatic brain injury, or TBI, is a disturbance of normal brain function induced by an external mechanical force, such as a direct blow to the head or a penetrating injury. It can affect both brain structure and function, producing a wide range of clinical outcomes. TBI is a heterogeneous condition, meaning its effects may differ based on the type, location, and severity of the injury.Basis of ClassificationTBI is classified based on severity, injury mechanism, or pathophysiology. In...