<p>Questions on the implementation and utility of artificial intelligence (AI) play an increasingly more important role in forensic age diagnostics of living persons, especially in forensic radiological approaches. Numerous studies are so far available on skeletal regions which are of interest for the age estimation. This study firstly provides a comprehensive overview of the whole topic and finally provides detailed examples of the application of hand radiography and orthopantomography. The considerations emphasize that convolutional neural networks (CNN) and deep convolutional neural networks (DCNN) have been proven to be very suitable. As most forensically relevant investigations on the use of AI in forensic age diagnostics are currently based on orthopantomography, more data on other skeletal regions appear to be desirable. Future developments should not only concentrate on the optimization of the models and the application of larger and more diverse datasets, but also on the creation of unified standards to increase the comparability of the results.</p>

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Künstliche Intelligenz in der forensisch-radiologischen Altersdiagnostik

  • Maria Hahnemann,
  • Andreas Heinrich,
  • Hans-Joachim Mentzel,
  • Daniel Wittschieber

摘要

Questions on the implementation and utility of artificial intelligence (AI) play an increasingly more important role in forensic age diagnostics of living persons, especially in forensic radiological approaches. Numerous studies are so far available on skeletal regions which are of interest for the age estimation. This study firstly provides a comprehensive overview of the whole topic and finally provides detailed examples of the application of hand radiography and orthopantomography. The considerations emphasize that convolutional neural networks (CNN) and deep convolutional neural networks (DCNN) have been proven to be very suitable. As most forensically relevant investigations on the use of AI in forensic age diagnostics are currently based on orthopantomography, more data on other skeletal regions appear to be desirable. Future developments should not only concentrate on the optimization of the models and the application of larger and more diverse datasets, but also on the creation of unified standards to increase the comparability of the results.