DMA‐Net: A dual branch encoder and multi‐scale cross attention fusion network for skin lesion segmentation

Abstract Automatic segmentation of skin lesion is an important step in computer‐aided diagnosis. However, due to the significant variations in the size and shape of the lesion areas, as well as the low contrast with normal skin tissue, the boundaries are not clearly distinguishable, leading to a hig...

Full description

Saved in:
Bibliographic Details
Main Authors: Guangyao Zhai, Guanglei Wang, Qinghua Shang, Yan Li, Hongrui Wang
Format: Article
Language:English
Published: Wiley 2024-12-01
Series:IET Image Processing
Subjects:
Online Access:https://doi.org/10.1049/ipr2.13265
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1846121443204530176
author Guangyao Zhai
Guanglei Wang
Qinghua Shang
Yan Li
Hongrui Wang
author_facet Guangyao Zhai
Guanglei Wang
Qinghua Shang
Yan Li
Hongrui Wang
author_sort Guangyao Zhai
collection DOAJ
description Abstract Automatic segmentation of skin lesion is an important step in computer‐aided diagnosis. However, due to the significant variations in the size and shape of the lesion areas, as well as the low contrast with normal skin tissue, the boundaries are not clearly distinguishable, leading to a high possibility of incorrect segmentation. Therefore, this task is highly challenging. To overcome these difficulties, this paper proposes a medical image segmentation architecture named dual branch encoder and multi‐scale cross attention fusion network, which includes a dual‐branch encoder based on convolutional neural network and an improved channel‐enhanced Mamba to comprehensively extract local and global information from dermoscopy images. Additionally, to enhance the feature interaction and fusion of local and global information, a multi‐scale cross attention fusion module is adopted to cross‐merge features in different directions and at different scales, maximizing the advantages of the dual‐branch encoder and achieving precise segmentation of skin lesions. Extensive experiments are conducted on three public skin lesion datasets: ISIC‐2018, ISIC‐2017, and ISIC‐2016, to verify the effectiveness and superiority of the proposed method. The dice similarity coefficient scores on the three datasets reached 81.77%, 81.68% and 85.60%, respectively, surpassing most state‐of‐the‐art methods.
format Article
id doaj-art-67d379026c1f40fcb2eb8d53f867b295
institution Kabale University
issn 1751-9659
1751-9667
language English
publishDate 2024-12-01
publisher Wiley
record_format Article
series IET Image Processing
spelling doaj-art-67d379026c1f40fcb2eb8d53f867b2952024-12-16T04:00:31ZengWileyIET Image Processing1751-96591751-96672024-12-0118144531454110.1049/ipr2.13265DMA‐Net: A dual branch encoder and multi‐scale cross attention fusion network for skin lesion segmentationGuangyao Zhai0Guanglei Wang1Qinghua Shang2Yan Li3Hongrui Wang4College of Electronic and Information Engineering Hebei University Hebei ChinaCollege of Electronic and Information Engineering Hebei University Hebei ChinaCollege of Electronic and Information Engineering Hebei University Hebei ChinaCollege of Electronic and Information Engineering Hebei University Hebei ChinaCollege of Electronic and Information Engineering Hebei University Hebei ChinaAbstract Automatic segmentation of skin lesion is an important step in computer‐aided diagnosis. However, due to the significant variations in the size and shape of the lesion areas, as well as the low contrast with normal skin tissue, the boundaries are not clearly distinguishable, leading to a high possibility of incorrect segmentation. Therefore, this task is highly challenging. To overcome these difficulties, this paper proposes a medical image segmentation architecture named dual branch encoder and multi‐scale cross attention fusion network, which includes a dual‐branch encoder based on convolutional neural network and an improved channel‐enhanced Mamba to comprehensively extract local and global information from dermoscopy images. Additionally, to enhance the feature interaction and fusion of local and global information, a multi‐scale cross attention fusion module is adopted to cross‐merge features in different directions and at different scales, maximizing the advantages of the dual‐branch encoder and achieving precise segmentation of skin lesions. Extensive experiments are conducted on three public skin lesion datasets: ISIC‐2018, ISIC‐2017, and ISIC‐2016, to verify the effectiveness and superiority of the proposed method. The dice similarity coefficient scores on the three datasets reached 81.77%, 81.68% and 85.60%, respectively, surpassing most state‐of‐the‐art methods.https://doi.org/10.1049/ipr2.13265codecsimage segmentationskin
spellingShingle Guangyao Zhai
Guanglei Wang
Qinghua Shang
Yan Li
Hongrui Wang
DMA‐Net: A dual branch encoder and multi‐scale cross attention fusion network for skin lesion segmentation
IET Image Processing
codecs
image segmentation
skin
title DMA‐Net: A dual branch encoder and multi‐scale cross attention fusion network for skin lesion segmentation
title_full DMA‐Net: A dual branch encoder and multi‐scale cross attention fusion network for skin lesion segmentation
title_fullStr DMA‐Net: A dual branch encoder and multi‐scale cross attention fusion network for skin lesion segmentation
title_full_unstemmed DMA‐Net: A dual branch encoder and multi‐scale cross attention fusion network for skin lesion segmentation
title_short DMA‐Net: A dual branch encoder and multi‐scale cross attention fusion network for skin lesion segmentation
title_sort dma net a dual branch encoder and multi scale cross attention fusion network for skin lesion segmentation
topic codecs
image segmentation
skin
url https://doi.org/10.1049/ipr2.13265
work_keys_str_mv AT guangyaozhai dmanetadualbranchencoderandmultiscalecrossattentionfusionnetworkforskinlesionsegmentation
AT guangleiwang dmanetadualbranchencoderandmultiscalecrossattentionfusionnetworkforskinlesionsegmentation
AT qinghuashang dmanetadualbranchencoderandmultiscalecrossattentionfusionnetworkforskinlesionsegmentation
AT yanli dmanetadualbranchencoderandmultiscalecrossattentionfusionnetworkforskinlesionsegmentation
AT hongruiwang dmanetadualbranchencoderandmultiscalecrossattentionfusionnetworkforskinlesionsegmentation