Multi-pathology detection and lesion localization in WCE videos by using the instance segmentation approach

作者：

Highlights：

• Method for segmentation and classification of multiple lesions in Capsule Endoscopy

• Novel Method of mask subnet scheme R-CNN for improvements in mask predictions

• Second order moment for network training, improving the results of single momentum

• Improvements of almost 7% in mAP when compared to PANet results

摘要

•Method for segmentation and classification of multiple lesions in Capsule Endoscopy•Novel Method of mask subnet scheme R-CNN for improvements in mask predictions•Second order moment for network training, improving the results of single momentum•Improvements of almost 7% in mAP when compared to PANet results

论文关键词：Multi-pathology detection,Lesion localization,Instance segmentation,Mask-RCNN,Wireless capsule endoscopy,PANet

论文评审过程：Received 5 January 2021, Revised 10 June 2021, Accepted 3 August 2021, Available online 10 August 2021, Version of Record 12 August 2021.

论文官网地址：https://doi.org/10.1016/j.artmed.2021.102141