From spoken narratives to domain knowledge: Mining linguistic data for medical image understanding

作者：

Highlights：

• The medical images that can be described more consensually by physicians tend to be less diagnostically challenging.

• The medical images that can be described with highly related concepts by most physicians tend to be less diagnostically challenging.

• We develop a method based on anchor concepts to cluster diagnostic narratives by image.

摘要

Highlights•The medical images that can be described more consensually by physicians tend to be less diagnostically challenging.•The medical images that can be described with highly related concepts by most physicians tend to be less diagnostically challenging.•We develop a method based on anchor concepts to cluster diagnostic narratives by image.

论文关键词：Unified Medical Language System,Lexical consensus,Semantic relatedness,Clustering algorithm,Image-based diagnostic reasoning,Medical data analysis

论文评审过程：Received 2 June 2013, Revised 29 July 2014, Accepted 10 August 2014, Available online 19 August 2014.

论文官网地址：https://doi.org/10.1016/j.artmed.2014.08.001