Multi-stage domain-specific pretraining for improved detection and localization of Barrett's neoplasia: A comprehensive clinically validated study

作者：

Highlights：

• State of the art results on dysplasia detection in Barrett's esophagus patients.

• Unique pre-training with a large gastrointestinal database (data is shareable).

• Evaluation on two test sets. One with general cases and one with hard cases.

• Compared to 53 medical professionals with 4 different skill.

• Results validated in a live in real-time on unseen patients.

摘要

•State of the art results on dysplasia detection in Barrett's esophagus patients.•Unique pre-training with a large gastrointestinal database (data is shareable).•Evaluation on two test sets. One with general cases and one with hard cases.•Compared to 53 medical professionals with 4 different skill.•Results validated in a live in real-time on unseen patients.

论文关键词：Computer-aided detection,Barrett's Esophagus,Deep learning,Clinical validation

论文评审过程：Received 8 January 2020, Revised 8 May 2020, Accepted 15 June 2020, Available online 18 June 2020, Version of Record 25 June 2020.

论文官网地址：https://doi.org/10.1016/j.artmed.2020.101914