Vi Sion Transformer and Large Language Model-Enhanced Generative Artificial Intelligence for Automatic Reporting of Bone Mineral Densitometry Using the Chest Computed Tomographic Radiomics Method

Project: A - Government Institutionb - National Science and Technology Council

Project Details

Key findings

骨質疏鬆症是全球人口老化重大的公共衛生挑戰,椎骨骨折可造成立即失能,產生極大的健保負擔。然而大多數慢性骨質流失的病人對於自己的骨質狀況一無所知,這個急迫的未滿足需求可藉由人工智慧模型在常規CT影像的PACS系統內進行「機會性篩檢」,可以提高早期發現骨質疏鬆的比率。新的影像技術,例如放射組學、視覺轉換器和大語言模型等先進的人工智慧方法能改變骨質疏鬆症的篩檢並提供了新的解決方案。我們的預期結果將可使椎骨骨質在標準胸部CT檢查期間在背景中進行例行機會性篩檢,可為臨床醫師提供精準的風險評估,並強化預防和治療決策。本計畫能實現醫用人工智慧賦能的精準診斷運用和臨床處置骨質疏鬆症的重大進步。
StatusActive
Effective start/end date8/1/247/31/27

Keywords

  • Osteoporosis
  • Bone Mineral Density
  • Opportunistic Screening
  • Preventive Medicine
  • Computed Tomography
  • Artificial Intelligence
  • Vision Transformer
  • Multilayer Perceptron
  • Bidirectional Long Short-Term Memory
  • Residual Neural Network
  • Image Autosegmentation
  • Radiomics
  • Large Language Model
  • Bidirectional Encoder Representations from Transformers
  • Natural Language Processing
  • Instruction Learning