Data Management And Sharing System of China National Clinical Research Center for Neurological Diseases

Large Foundation Model based Deep Learning System to Predict Recurrence and Disability Outcomes in Patients with Transient Ischemic Attack or Ischemic Stroke

First author:Ziyang Liu

Corresponding author:Tao Liu,Yongjun Wang

essential_information

  • Person in charge:Ziyang Liu
  • Data Sources: CNSR-Ⅲ
  • Number of papers:CNSR-Ⅲ20230092
  • Published time:Invalid da

Invalid da

Published articles

Journal:

Publication type: Original Contributions

IF(IF):

PMID: Full text Download

Background And Purpose

本项目聚焦于脑血管病的神经影像分析,建立j基于大模型的深度学习模型实现1)对脑血管病的临床结局(包括复发,死亡,致残)以及认知能力的预测;2)以数据驱动模式自动标记影像中潜在的病灶区域;

Method

基于预训练大模型,针对脑血管病的临床结局,采用Fine Tuning Method & Prompt Engineering 方法进行模型的迁移学习训练。 输入:多模态影像数据;全图像输入/Patch图像输入 模型:建立多模态多任务(分类和回归)的深度学习预测模型,从神经影像中挖掘高层次特征,对临床结局和认知能力进行准确预测。 结果:复发,死亡,致残(离散型)等临床结局的准确率,以及认知分数(连续型)与真实值的差异指标。

Conclusion

Main results