Integration of Human and Machine Observers for the Interpretation of Medical Images

整合人类和机器观察者来解读医学图像

• 批准号: 516276-2017
• 负责人: Klein, Ran
• 金额: $ 1.82万
• 依托单位: University of Ottawa
• 依托单位国家: 加拿大
• 项目类别: Engage Grants Program
• 财政年份: 2017
• 资助国家: 加拿大
• 起止时间: 2017-01-01 至 2018-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=639929
• 关键词: Integration; Human; Machine; Observers; Interpretation

Medical images must be interpreted by medical professionals to determine the presence of disease. Like allhumans, medical professionals have limited attention and make mistakes. Furthermore, modern imagingequipment produce larger and more complex data which the human observer struggles to analyze. Moderncomputer algorithms can identify more subtle patterns, can analyze larger datasets, learn from more examplesand are reproducible. Nevertheless we do not always know how to program computers for certain tasks andtherefore rely on adaptive algorithms to "learn" to detect disease. To leverage the strength of expert humanobservers and machine observers we require two-way collaboration between man and machine. In this projectwe propose to adapt an existing medical image viewing system, in collaboration with its vendor (HermesMedical Solutions), to enable computer aided diagnosis (where the machine helps the physician) andsupervised machine learning (where the user provides corrective feedback to the learning machine).

医学图像必须由医学专业人员解释，以确定疾病的存在。像所有人一样，医疗专业人员的注意力有限，也会犯错误。此外，现代成像设备产生更大和更复杂的数据，人类观察者难以分析。现代计算机算法可以识别更微妙的模式，可以分析更大的数据集，从更多的例子中学习，并且是可重复的。然而，我们并不总是知道如何为计算机编程以完成某些任务，因此我们依赖自适应算法来“学习”检测疾病。为了充分利用专家人类观察者和机器观察者的力量，我们需要人与机器之间的双向合作。在这个项目中，我们建议与其供应商（Hermes Medical Solutions）合作，调整现有的医学图像查看系统，以实现计算机辅助诊断（机器帮助医生）和监督机器学习（用户向学习机器提供纠正反馈）。