Evolutionary deep feature selection for compact representation of gigapixel images in digital pathology
作者：
Highlights：
• Deep learning increasingly influences in digital pathology workflow.
• Compactly representing a WSI to enable computational pathology is an urgent need.
• Evolutionary computation can optimize the output of pre-trained deep networks.
• Irrelevant or redundant features are removed to encompass salient features.
• Compact feature vectors achieved 93% classification accuracy.
摘要
•Deep learning increasingly influences in digital pathology workflow.•Compactly representing a WSI to enable computational pathology is an urgent need.•Evolutionary computation can optimize the output of pre-trained deep networks.•Irrelevant or redundant features are removed to encompass salient features.•Compact feature vectors achieved 93% classification accuracy.
论文关键词：Digital pathology,Whole slide images,Image representation,Evolutionary computation
论文评审过程：Received 26 September 2021, Revised 13 June 2022, Accepted 14 July 2022, Available online 25 July 2022, Version of Record 2 August 2022.
论文官网地址：https://doi.org/10.1016/j.artmed.2022.102368
