Class-Agnostic Feature-Learning-Based Deep-Learning Model for Robust Melanoma Prediction

Yuseong Chu; Solam Lee; Byungho Oh; Sejung Yang

doi:Skin lesion images are relatively easy to obtain, but variations in imaging conditions, particularly lesion positioning, can cause inconsistencies in deep learning model activations, impacting diagnostic reliability. In this study, a robust deep-learning model for melanoma prediction was developed using class-agnostic activation maps (CAAMs) for enhanced diagnostic accuracy and reliability by addressing issues related to image variability and transformation robustness. The international skin imaging collaboration (ISIC) 2017 and 2019 datasets, focusing on melanoma and nevus, were used. The performance was evaluated using the area under the receiver operating characteristic curve (AUROC), while Dice scores were used for robustness evaluation. The proposed model achieved an AUROC of 0.954 for ConvNeXt on the ISIC 2019 dataset, with Dice scores of 0.664 and 0.457 for ConvNeXt and ResNet, respectively. For the ISIC 2017 dataset, the proposed model achieved an AUROC of 0.843 for ConvNeXt, with Dice scores of 0.557 and 0.306 for ConvNeXt and ResNet, respectively. The proposed CAAM-based method improved the melanoma prediction and lesion recognition accuracy by ensuring robust and consistent CAAMs.

YUHSpace

BROWSE

0 0

Cited 0 times in

Class-Agnostic Feature-Learning-Based Deep-Learning Model for Robust Melanoma Prediction

DC Field	Value	Language
dc.contributor.author	오병호	-
dc.date.accessioned	2025-10-15T01:33:36Z	-
dc.date.available	2025-10-15T01:33:36Z	-
dc.date.issued	2025-07	-
dc.identifier.issn	2168-2194	-
dc.identifier.uri	https://ir.ymlib.yonsei.ac.kr/handle/22282913/207408	-
dc.description.statementOfResponsibility	https://ieeexplore-ieee-org-ssl.access.yonsei.ac.kr/document/10856449	-
dc.language	English	-
dc.publisher	Institute of Electrical and Electronics Engineers	-
dc.relation.isPartOf	IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS	-
dc.rights	CC BY-NC-ND 2.0 KR	-
dc.subject.MESH	1	-
dc.title	Class-Agnostic Feature-Learning-Based Deep-Learning Model for Robust Melanoma Prediction	-
dc.type	Article	-
dc.contributor.college	College of Medicine (의과대학)	-
dc.contributor.department	Dept. of Dermatology (피부과학교실)	-
dc.contributor.googleauthor	Yuseong Chu	-
dc.contributor.googleauthor	Solam Lee	-
dc.contributor.googleauthor	Byungho Oh	-
dc.contributor.googleauthor	Sejung Yang	-
dc.identifier.doi	Skin lesion images are relatively easy to obtain, but variations in imaging conditions, particularly lesion positioning, can cause inconsistencies in deep learning model activations, impacting diagnostic reliability. In this study, a robust deep-learning model for melanoma prediction was developed using class-agnostic activation maps (CAAMs) for enhanced diagnostic accuracy and reliability by addressing issues related to image variability and transformation robustness. The international skin imaging collaboration (ISIC) 2017 and 2019 datasets, focusing on melanoma and nevus, were used. The performance was evaluated using the area under the receiver operating characteristic curve (AUROC), while Dice scores were used for robustness evaluation. The proposed model achieved an AUROC of 0.954 for ConvNeXt on the ISIC 2019 dataset, with Dice scores of 0.664 and 0.457 for ConvNeXt and ResNet, respectively. For the ISIC 2017 dataset, the proposed model achieved an AUROC of 0.843 for ConvNeXt, with Dice scores of 0.557 and 0.306 for ConvNeXt and ResNet, respectively. The proposed CAAM-based method improved the melanoma prediction and lesion recognition accuracy by ensuring robust and consistent CAAMs.	-
dc.contributor.localId	A02367	-
dc.relation.journalcode	J03267	-
dc.identifier.eissn	2168-2208	-
dc.identifier.pmid	40031346	-
dc.subject.keyword	Databases, Factual	-
dc.subject.keyword	Deep Learning*	-
dc.subject.keyword	Humans	-
dc.subject.keyword	Image Interpretation, Computer-Assisted* / methods	-
dc.subject.keyword	Melanoma* / diagnostic imaging	-
dc.subject.keyword	ROC Curve	-
dc.subject.keyword	Skin / diagnostic imaging	-
dc.subject.keyword	Skin Neoplasms* / diagnostic imaging	-
dc.contributor.alternativeName	Oh, Byung Ho	-
dc.contributor.affiliatedAuthor	오병호	-
dc.citation.volume	29	-
dc.citation.number	7	-
dc.citation.startPage	4946	-
dc.citation.endPage	4955	-
dc.identifier.bibliographicCitation	IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS, Vol.29(7) : 4946-4955, 2025-07	-
dc.identifier.articleno	10.1109/JBHI.2025.3535536	-

Appears in Collections:: 1. College of Medicine (의과대학) > Dept. of Dermatology (피부과학교실) > 1. Journal Papers

Show simple item record Find it @ YMLIB

License

YUHSpace: Class-Agnostic Feature-Learning-Based Deep-Learning Model for Robust Melanoma Prediction

YUHSpace

BROWSE

Browse

Links