IARS SegNet: Interpretable Attention Residual Skip Connection SegNet for Melanoma Segmentation

Skin lesion segmentation plays a crucial role in the computer-aided diagnosis of melanoma. Deep Learning models have shown promise in accurately segmenting skin lesions, but their widespread adoption in real-life clinical settings is hindered by their inherent black-box nature. In domains as critica...

ver descrição completa

Detalhes bibliográficos
Autores: Narayanan, V. Shankara, Sikha, O. K., Benitez, Raul
Tipo de documento: artigo
Estado:Versão publicada
Data de publicação:2024
País:España
Recursos:Fundació Sant Joan de Déu
Repositório:r-FSJD. Repositorio Institucional de Producción Científica de la Fundació Sant Joan de Déu
OAI Identifier:oai:fsjd.fundanetsuite.com:p27458
Acesso em linha:https://fsjd.fundanetsuite.com/Publicaciones/ProdCientif/PublicacionFrw.aspx?id=27458
Access Level:Acceso aberto
Palavra-chave:Lesions
Skin
Computer architecture
Image segmentation
Melanoma
Computational modeling
Feature extraction
Semantic segmentation
Explainable AI
Deep learning
explainable AI
skin lesion segmentation
deep learning
id ES_e1de59e87ea280cc2d7bc9eb1da3b1c2
oai_identifier_str oai:fsjd.fundanetsuite.com:p27458
network_acronym_str ES
network_name_str España
repository_id_str
spelling IARS SegNet: Interpretable Attention Residual Skip Connection SegNet for Melanoma SegmentationNarayanan, V. ShankaraSikha, O. K.Benitez, RaulLesionsSkinComputer architectureImage segmentationMelanomaComputational modelingFeature extractionSemantic segmentationExplainable AIDeep learningexplainable AIskin lesion segmentationdeep learningSkin lesion segmentation plays a crucial role in the computer-aided diagnosis of melanoma. Deep Learning models have shown promise in accurately segmenting skin lesions, but their widespread adoption in real-life clinical settings is hindered by their inherent black-box nature. In domains as critical as healthcare, interpretability is not merely a feature but a fundamental requirement for model adoption. This paper proposes IARS SegNet an advanced segmentation framework built upon the SegNet baseline model. Our approach incorporates three critical components: Skip connections, residual convolutions, and a global attention mechanism onto the baseline Segnet architecture. These elements play a pivotal role in accentuating the significance of clinically relevant regions, particularly the contours of skin lesions. The inclusion of skip connections enhances the model's capacity to learn intricate contour details, while the use of residual convolutions allows for the construction of a deeper model while preserving essential image features. The global attention mechanism further contributes by extracting refined feature maps from each convolutional and deconvolutional block, thereby elevating the model's interpretability. This enhancement highlights critical regions, fosters better understanding, and leads to more accurate skin lesion segmentation for melanoma diagnosis. This study primarily focuses on the interpretation of performance improvements in the base model resulting from the integration of each of these three components. To comprehensively assess the performance gain achieved with each addition, we employ two sets of evaluation metrics, quantifying performance based on both regions and contours. The results underscore the superior segmentation capabilities of the proposed architecture compared to the SegNet and U-Net models. Notably, the model's performance gain due to each enhancement is visually interpretable.IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC2024info:eu-repo/semantics/articleinfo:eu-repo/semantics/publishedVersionhttps://fsjd.fundanetsuite.com/Publicaciones/ProdCientif/PublicacionFrw.aspx?id=27458IEEE AccessISSN: 21693536reponame:r-FSJD. Repositorio Institucional de Producción Científica de la Fundació Sant Joan de Déuinstname:Fundació Sant Joan de DéuInglésinfo:eu-repo/semantics/openAccessoai:fsjd.fundanetsuite.com:p274582026-05-27T12:37:41Z
dc.title.none.fl_str_mv IARS SegNet: Interpretable Attention Residual Skip Connection SegNet for Melanoma Segmentation
title IARS SegNet: Interpretable Attention Residual Skip Connection SegNet for Melanoma Segmentation
spellingShingle IARS SegNet: Interpretable Attention Residual Skip Connection SegNet for Melanoma Segmentation
Narayanan, V. Shankara
Lesions
Skin
Computer architecture
Image segmentation
Melanoma
Computational modeling
Feature extraction
Semantic segmentation
Explainable AI
Deep learning
explainable AI
skin lesion segmentation
deep learning
title_short IARS SegNet: Interpretable Attention Residual Skip Connection SegNet for Melanoma Segmentation
title_full IARS SegNet: Interpretable Attention Residual Skip Connection SegNet for Melanoma Segmentation
title_fullStr IARS SegNet: Interpretable Attention Residual Skip Connection SegNet for Melanoma Segmentation
title_full_unstemmed IARS SegNet: Interpretable Attention Residual Skip Connection SegNet for Melanoma Segmentation
title_sort IARS SegNet: Interpretable Attention Residual Skip Connection SegNet for Melanoma Segmentation
dc.creator.none.fl_str_mv Narayanan, V. Shankara
Sikha, O. K.
Benitez, Raul
author Narayanan, V. Shankara
author_facet Narayanan, V. Shankara
Sikha, O. K.
Benitez, Raul
author_role author
author2 Sikha, O. K.
Benitez, Raul
author2_role author
author
dc.subject.none.fl_str_mv Lesions
Skin
Computer architecture
Image segmentation
Melanoma
Computational modeling
Feature extraction
Semantic segmentation
Explainable AI
Deep learning
explainable AI
skin lesion segmentation
deep learning
topic Lesions
Skin
Computer architecture
Image segmentation
Melanoma
Computational modeling
Feature extraction
Semantic segmentation
Explainable AI
Deep learning
explainable AI
skin lesion segmentation
deep learning
description Skin lesion segmentation plays a crucial role in the computer-aided diagnosis of melanoma. Deep Learning models have shown promise in accurately segmenting skin lesions, but their widespread adoption in real-life clinical settings is hindered by their inherent black-box nature. In domains as critical as healthcare, interpretability is not merely a feature but a fundamental requirement for model adoption. This paper proposes IARS SegNet an advanced segmentation framework built upon the SegNet baseline model. Our approach incorporates three critical components: Skip connections, residual convolutions, and a global attention mechanism onto the baseline Segnet architecture. These elements play a pivotal role in accentuating the significance of clinically relevant regions, particularly the contours of skin lesions. The inclusion of skip connections enhances the model's capacity to learn intricate contour details, while the use of residual convolutions allows for the construction of a deeper model while preserving essential image features. The global attention mechanism further contributes by extracting refined feature maps from each convolutional and deconvolutional block, thereby elevating the model's interpretability. This enhancement highlights critical regions, fosters better understanding, and leads to more accurate skin lesion segmentation for melanoma diagnosis. This study primarily focuses on the interpretation of performance improvements in the base model resulting from the integration of each of these three components. To comprehensively assess the performance gain achieved with each addition, we employ two sets of evaluation metrics, quantifying performance based on both regions and contours. The results underscore the superior segmentation capabilities of the proposed architecture compared to the SegNet and U-Net models. Notably, the model's performance gain due to each enhancement is visually interpretable.
publishDate 2024
dc.date.none.fl_str_mv 2024
dc.type.none.fl_str_mv info:eu-repo/semantics/article
info:eu-repo/semantics/publishedVersion
format article
status_str publishedVersion
dc.identifier.none.fl_str_mv https://fsjd.fundanetsuite.com/Publicaciones/ProdCientif/PublicacionFrw.aspx?id=27458
url https://fsjd.fundanetsuite.com/Publicaciones/ProdCientif/PublicacionFrw.aspx?id=27458
dc.language.none.fl_str_mv Inglés
language_invalid_str_mv Inglés
dc.rights.none.fl_str_mv info:eu-repo/semantics/openAccess
eu_rights_str_mv openAccess
dc.publisher.none.fl_str_mv IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
publisher.none.fl_str_mv IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
dc.source.none.fl_str_mv IEEE Access
ISSN: 21693536
reponame:r-FSJD. Repositorio Institucional de Producción Científica de la Fundació Sant Joan de Déu
instname:Fundació Sant Joan de Déu
instname_str Fundació Sant Joan de Déu
reponame_str r-FSJD. Repositorio Institucional de Producción Científica de la Fundació Sant Joan de Déu
collection r-FSJD. Repositorio Institucional de Producción Científica de la Fundació Sant Joan de Déu
repository.name.fl_str_mv
repository.mail.fl_str_mv
_version_ 1869422333706371072
score 15,812429

Registros relacionados