Slide
Centre Interdisciplinaire
de Recherche et d’Innovation
en Cybersécurité et Société
de Recherche et d’Innovation
en Cybersécurité et Société
1.
Allaoui, M. L.; Allili, M. S.
MEDiXNet: A Robust Mixture of Expert Dermatological Imaging Networks for Skin Lesion Segmentation Article d'actes
Dans: IEEE Comput. Soc. Conf. Comput. Vis. Pattern Recogn., IEEE Computer Society, 2024, ISBN: 19457928 (ISSN); 979-835031333-8 (ISBN), (Journal Abbreviation: IEEE Comput. Soc. Conf. Comput. Vis. Pattern Recogn.).
Résumé | Liens | BibTeX | Étiquettes: Attention mechanism, Attention mechanisms, Blurred boundaries, Cancer detection, Deep learning, Dermatology, Expert systems, Image segmentation, Lesion segmentations, Mixture of experts, Mixture of experts model, Mixture-of-experts model, Salient regions, Skin cancers, Skin lesion, Skin lesion segmentation
@inproceedings{allaoui_medixnet_2024,
title = {MEDiXNet: A Robust Mixture of Expert Dermatological Imaging Networks for Skin Lesion Segmentation},
author = {M. L. Allaoui and M. S. Allili},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85203397643&doi=10.1109%2fISBI56570.2024.10635430&partnerID=40&md5=c95dd2122f03c944e945b684a111e741},
doi = {10.1109/ISBI56570.2024.10635430},
isbn = {19457928 (ISSN); 979-835031333-8 (ISBN)},
year = {2024},
date = {2024-01-01},
booktitle = {IEEE Comput. Soc. Conf. Comput. Vis. Pattern Recogn.},
publisher = {IEEE Computer Society},
abstract = {Skin lesion segmentation in dermatological imaging is crucial for early skin cancer detection. However, it is challenging due to variation in lesion appearance, blurred boundaries, and the presence of artifacts. Existing segmentation methods often fall short in accurately addressing these issues. We present MEDiXNet, a novel deep learning model combining expert networks with the Adaptive Salient Region Attention Module (ASRAM) to specifically tackle these challenges. Tailored for varying lesion types, MEDiXNet leverages ASRAM to enhance focus on critical regions, substantially improving segmentation accuracy. Tested on the ISIC datasets, it achieved a 94% Dice coefficient, surpassing state-of-the-art methods. MEDiXNet's innovative approach represents a significant advancement in dermatological imaging, promising to elevate the precision of skin cancer diagnostics. © 2024 IEEE.},
note = {Journal Abbreviation: IEEE Comput. Soc. Conf. Comput. Vis. Pattern Recogn.},
keywords = {Attention mechanism, Attention mechanisms, Blurred boundaries, Cancer detection, Deep learning, Dermatology, Expert systems, Image segmentation, Lesion segmentations, Mixture of experts, Mixture of experts model, Mixture-of-experts model, Salient regions, Skin cancers, Skin lesion, Skin lesion segmentation},
pubstate = {published},
tppubtype = {inproceedings}
}
Skin lesion segmentation in dermatological imaging is crucial for early skin cancer detection. However, it is challenging due to variation in lesion appearance, blurred boundaries, and the presence of artifacts. Existing segmentation methods often fall short in accurately addressing these issues. We present MEDiXNet, a novel deep learning model combining expert networks with the Adaptive Salient Region Attention Module (ASRAM) to specifically tackle these challenges. Tailored for varying lesion types, MEDiXNet leverages ASRAM to enhance focus on critical regions, substantially improving segmentation accuracy. Tested on the ISIC datasets, it achieved a 94% Dice coefficient, surpassing state-of-the-art methods. MEDiXNet's innovative approach represents a significant advancement in dermatological imaging, promising to elevate the precision of skin cancer diagnostics. © 2024 IEEE.
2.
Nouboukpo, A.; Allaoui, M. L.; Allili, M. S.
Multi-scale spatial consistency for deep semi-supervised skin lesion segmentation Article de journal
Dans: Engineering Applications of Artificial Intelligence, vol. 135, 2024, ISSN: 09521976 (ISSN), (Publisher: Elsevier Ltd).
Résumé | Liens | BibTeX | Étiquettes: Deep learning, Dermatology, Image segmentation, Lesion segmentations, Medical imaging, Multi-scales, Semi-supervised, Semi-supervised learning, Skin lesion, Skin lesion segmentation, Spatial consistency, Spatially constrained mixture model, Spatially-constrained mixture models, Supervised learning, UNets, Unlabeled data
@article{nouboukpo_multi-scale_2024,
title = {Multi-scale spatial consistency for deep semi-supervised skin lesion segmentation},
author = {A. Nouboukpo and M. L. Allaoui and M. S. Allili},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85195700182&doi=10.1016%2fj.engappai.2024.108681&partnerID=40&md5=e1cc2b6a1bb0aed530e8c04583c76167},
doi = {10.1016/j.engappai.2024.108681},
issn = {09521976 (ISSN)},
year = {2024},
date = {2024-01-01},
journal = {Engineering Applications of Artificial Intelligence},
volume = {135},
abstract = {This paper introduces a novel semi-supervised framework, the Multiscale Spatial Consistency Network (MSCNet), for robust semi-supervised skin lesion segmentation. MSCNet uses local and global spatial consistency to leverage a minimal set of labeled data, supplemented by a large number of unlabeled data, to improve segmentation. The model is is based on a single Encoder–Decoder (ED) network, augmented with a Spatially-Constrained Mixture Model (SCMM) to enforce spatial coherence in predictions. To encode the local spatial consistency, a hierarchical superpixel structure is used capture local region context (LRC), bolstering the model capacity to discern fine-grained lesion details. Global consistency is enforced through the SCMM module, which uses a larger context for lesion/background discrimination. In addition, it enables efficient leveraging of the unlabeled data through pseudo-label generation. Experiments demonstrate that the MSCNet outperforms existing state-of-the-art methods in segmenting complex lesions. The MSCNet has an excellent generalization capability, offering a promising direction for semi-supervised medical image segmentation, particularly in scenarios with limited annotated data. The code is available at https://github.com/AdamaTG/MSCNet. © 2024 Elsevier Ltd},
note = {Publisher: Elsevier Ltd},
keywords = {Deep learning, Dermatology, Image segmentation, Lesion segmentations, Medical imaging, Multi-scales, Semi-supervised, Semi-supervised learning, Skin lesion, Skin lesion segmentation, Spatial consistency, Spatially constrained mixture model, Spatially-constrained mixture models, Supervised learning, UNets, Unlabeled data},
pubstate = {published},
tppubtype = {article}
}
This paper introduces a novel semi-supervised framework, the Multiscale Spatial Consistency Network (MSCNet), for robust semi-supervised skin lesion segmentation. MSCNet uses local and global spatial consistency to leverage a minimal set of labeled data, supplemented by a large number of unlabeled data, to improve segmentation. The model is is based on a single Encoder–Decoder (ED) network, augmented with a Spatially-Constrained Mixture Model (SCMM) to enforce spatial coherence in predictions. To encode the local spatial consistency, a hierarchical superpixel structure is used capture local region context (LRC), bolstering the model capacity to discern fine-grained lesion details. Global consistency is enforced through the SCMM module, which uses a larger context for lesion/background discrimination. In addition, it enables efficient leveraging of the unlabeled data through pseudo-label generation. Experiments demonstrate that the MSCNet outperforms existing state-of-the-art methods in segmenting complex lesions. The MSCNet has an excellent generalization capability, offering a promising direction for semi-supervised medical image segmentation, particularly in scenarios with limited annotated data. The code is available at https://github.com/AdamaTG/MSCNet. © 2024 Elsevier Ltd