Honeycomb lung segmentation network based on P2T with CNN two-branch parallelism

Aiming at the problem that honeycomb lung lesions are difficult to accurately segment due to diverse morphology and complex distribution, a network with parallel two-branch structure is proposed. In the encoder, the Pyramid Pooling Transformer (P2T) backbone is used as the Transformer branch to obta...

Full description

Saved in:
Bibliographic Details
Main Authors: Zhichao Li, Gang Li, Ling Zhang, Guijuan Cheng, Shan Wu
Format: Article
Language:English
Published: Tsinghua University Press 2024-12-01
Series:Intelligent and Converged Networks
Subjects:
Online Access:https://www.sciopen.com/article/10.23919/ICN.2024.0023
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850091377492230144
author Zhichao Li
Gang Li
Ling Zhang
Guijuan Cheng
Shan Wu
author_facet Zhichao Li
Gang Li
Ling Zhang
Guijuan Cheng
Shan Wu
author_sort Zhichao Li
collection DOAJ
description Aiming at the problem that honeycomb lung lesions are difficult to accurately segment due to diverse morphology and complex distribution, a network with parallel two-branch structure is proposed. In the encoder, the Pyramid Pooling Transformer (P2T) backbone is used as the Transformer branch to obtain the global features of the lesions, the convolutional branch is used to extract the lesions’ local feature information, and the feature fusion module is designed to effectively fuse the features in the dual branches; subsequently, in the decoder, the channel prior convolutional attention is used to enhance the localization ability of the model to the lesion region. To resolve the problem of model accuracy degradation caused by the class imbalance of the dataset, an adaptive weighted hybrid loss function is designed for model training. Finally, extensive experimental results show that the method in this paper performs well on the Honeycomb Lung Dataset, with Intersection over Union (IoU), mean Intersection over Union (mIoU), Dice coefficient, and Precision (Pre) of 0.8750, 0.9363, 0.9298, and 0.9012, respectively, which are better than other methods. In addition, its IoU and Dice coefficient of 0.7941 and 0.8875 on the Covid dataset further prove its excellent performance.
format Article
id doaj-art-79d1e8134ce84863bced9072ddab89d0
institution DOAJ
issn 2708-6240
language English
publishDate 2024-12-01
publisher Tsinghua University Press
record_format Article
series Intelligent and Converged Networks
spelling doaj-art-79d1e8134ce84863bced9072ddab89d02025-08-20T02:42:24ZengTsinghua University PressIntelligent and Converged Networks2708-62402024-12-015433635510.23919/ICN.2024.0023Honeycomb lung segmentation network based on P2T with CNN two-branch parallelismZhichao Li0Gang Li1Ling Zhang2Guijuan Cheng3Shan Wu4College of Software, Taiyuan University of Technology, Taiyuan 030024, ChinaCollege of Software, Taiyuan University of Technology, Taiyuan 030024, ChinaCollege of Software, Taiyuan University of Technology, Taiyuan 030024, ChinaCollege of Software, Taiyuan University of Technology, Taiyuan 030024, ChinaCT Div, Shanxi Bethune Hospital, Taiyuan 030024, ChinaAiming at the problem that honeycomb lung lesions are difficult to accurately segment due to diverse morphology and complex distribution, a network with parallel two-branch structure is proposed. In the encoder, the Pyramid Pooling Transformer (P2T) backbone is used as the Transformer branch to obtain the global features of the lesions, the convolutional branch is used to extract the lesions’ local feature information, and the feature fusion module is designed to effectively fuse the features in the dual branches; subsequently, in the decoder, the channel prior convolutional attention is used to enhance the localization ability of the model to the lesion region. To resolve the problem of model accuracy degradation caused by the class imbalance of the dataset, an adaptive weighted hybrid loss function is designed for model training. Finally, extensive experimental results show that the method in this paper performs well on the Honeycomb Lung Dataset, with Intersection over Union (IoU), mean Intersection over Union (mIoU), Dice coefficient, and Precision (Pre) of 0.8750, 0.9363, 0.9298, and 0.9012, respectively, which are better than other methods. In addition, its IoU and Dice coefficient of 0.7941 and 0.8875 on the Covid dataset further prove its excellent performance.https://www.sciopen.com/article/10.23919/ICN.2024.0023honeycomb lung segmentationparallel two-branch architectureglobal-local feature integrationchannel prior convolutional attention
spellingShingle Zhichao Li
Gang Li
Ling Zhang
Guijuan Cheng
Shan Wu
Honeycomb lung segmentation network based on P2T with CNN two-branch parallelism
Intelligent and Converged Networks
honeycomb lung segmentation
parallel two-branch architecture
global-local feature integration
channel prior convolutional attention
title Honeycomb lung segmentation network based on P2T with CNN two-branch parallelism
title_full Honeycomb lung segmentation network based on P2T with CNN two-branch parallelism
title_fullStr Honeycomb lung segmentation network based on P2T with CNN two-branch parallelism
title_full_unstemmed Honeycomb lung segmentation network based on P2T with CNN two-branch parallelism
title_short Honeycomb lung segmentation network based on P2T with CNN two-branch parallelism
title_sort honeycomb lung segmentation network based on p2t with cnn two branch parallelism
topic honeycomb lung segmentation
parallel two-branch architecture
global-local feature integration
channel prior convolutional attention
url https://www.sciopen.com/article/10.23919/ICN.2024.0023
work_keys_str_mv AT zhichaoli honeycomblungsegmentationnetworkbasedonp2twithcnntwobranchparallelism
AT gangli honeycomblungsegmentationnetworkbasedonp2twithcnntwobranchparallelism
AT lingzhang honeycomblungsegmentationnetworkbasedonp2twithcnntwobranchparallelism
AT guijuancheng honeycomblungsegmentationnetworkbasedonp2twithcnntwobranchparallelism
AT shanwu honeycomblungsegmentationnetworkbasedonp2twithcnntwobranchparallelism