Multi-feature based benchmark for cervical dysplasia classification evaluation

作者：

Highlights：

• A new image dataset is introduced for evaluating cervical disease classification.

• The pyramid feature descriptor encodes complementary hand-crafted features.

• The deep convolutional neural network automatically learns features from images.

• Results using seven classic classifiers serve as the baseline for future research.

摘要

Highlights•A new image dataset is introduced for evaluating cervical disease classification.•The pyramid feature descriptor encodes complementary hand-crafted features.•The deep convolutional neural network automatically learns features from images.•Results using seven classic classifiers serve as the baseline for future research.

论文关键词：Cervical cancer screening,Computer aided diagnosis,Image classification,Pyramid histogram,Local binary patterns,Convolutional neural network

论文评审过程：Received 31 January 2016, Revised 20 June 2016, Accepted 21 September 2016, Available online 22 September 2016, Version of Record 27 November 2016.

论文官网地址：https://doi.org/10.1016/j.patcog.2016.09.027