常见间歇性肺炎的深度学习分类预测结果
Stephen M Humphries1, Devlin Thieke1, David Baraghoshi2
1Department of Radiology.
American journal of respiratory and critical care medicine
|January 11, 2024
概括
使用多个实例学习 (MIL) 的新型深度学习算法从CT扫描中准确预测通常的间歇性肺炎 (UIP). 这种人工智能工具增强了诊断信心,并有助于更早,更精确地识别间歇性肺病.
科学领域:
- 放射学和医学成像学 医学成像学
- 人工智能在医学中的应用
- 肺部医学 肺部医学
背景情况:
- 计算机断层扫描 (CT) 对于非侵入性常见间歇性肺炎 (UIP) 诊断至关重要.
- 由于CT扫描视觉评估的局限性,需要改进图像分析技术.
- 提高诊断准确度对于管理间歇性肺部疾病至关重要.
研究的目的:
- 开发一种可解释的深度学习算法,使用多个实例学习 (MIL) 来从CT进行UIP预测.
- 在独立的临床队列中验证MIL算法的性能.
- 评估算法的预测患者生存和肺功能下降的能力.
主要方法:
- 一个MIL算法被训练在一个聚合的数据集 (n=2,143) 和验证在三个独立的队列 (n=127,n=239,n=979).
- 使用接受器操作特征分析与组织学UIP作为基本事实来评估性能.
- 考克斯的比例危险和线性混合效应模型分析了MIL预测,死亡率和强迫生命能力 (FVC) 减少之间的关联.
主要成果:
- 与视觉评估相比,MIL算法在两个队列中显示出组织学UIP分类的准确性有所提高 (AUC为0.77和0.79与0.65和0.71对比).
- 在独立队列中,MIL-UIP分类是死亡率的显著预测指标 (未调整的HR为3.1和3.64,P<0.001).
- 根据MIL分类为UIP阳性的患者,经过纤维化程度调整后,每年FVC下降幅度显著增加 (-88毫升/年 vs. -45毫升/年,P<0.01).
结论:
- 使用MIL的计算机评估有效地识别了CT上临床意义上的UIP特征.
- 这种人工智能驱动的方法可以提高对间歇性肺部疾病的放射性评估的信心.
- 这种MIL方法有可能更早,更准确地诊断UIP,从而改善患者管理.
相关概念视频
Pneumonia I: Introduction
239
Pneumonia is an acute respiratory infection that targets the lungs, specifically the alveoli. These tiny air sacs, essential for oxygen exchange, become engorged with pus and fluid, severely hindering breathing, decreasing oxygen absorption, and causing significant pain and discomfort during respiration.
Risk Factors
Various factors influence the likelihood of developing pneumonia. Age plays a crucial role, with infants, children under two, and individuals over 65 at increased risk due to their...
Risk Factors
Various factors influence the likelihood of developing pneumonia. Age plays a crucial role, with infants, children under two, and individuals over 65 at increased risk due to their...
239
Pneumonia III: Complications and Assessment
227
Pneumonia poses the potential for numerous complications that warrant consideration. These complications include the following:
227
Pneumonia IV: Management
331
The treatment of pneumonia varies based on its severity and the causative pathogen. Here is a structured approach to managing pneumonia, integrating pharmaceutical and supportive care strategies.
Bacterial Pneumonia Treatment
For bacterial pneumonia, antibiotics serve as the cornerstone of therapy. Initial treatment often begins with empirical antibiotics, tailored to the anticipated causative organism and adjusted based on culture results. Key antibiotic choices include:
Bacterial Pneumonia Treatment
For bacterial pneumonia, antibiotics serve as the cornerstone of therapy. Initial treatment often begins with empirical antibiotics, tailored to the anticipated causative organism and adjusted based on culture results. Key antibiotic choices include:
331


