HEVA: Integrating ViT and ResNet50 with ABCD Clinical Features for Skin Cancer Diagnosis

Antônio Marcio Crepaldi Junior; Bianca Aparecida Andrade; Ademar Takeo Akabane

doi:10.22456/2175-2745.150917

Authors

Antônio Marcio Crepaldi Junior Centro de Pesquisa e Desenvolvimento em Telecomunicações (CPQD) https://orcid.org/0009-0000-8121-8105
Bianca Aparecida Andrade Pontifícia Universidade Católica de Campinas (PUC-Campinas)
Ademar Takeo Akabane Pontifícia Universidade Católica de Campinas (PUC-Campinas) https://orcid.org/0000-0002-4930-182X

DOI:

https://doi.org/10.22456/2175-2745.150917

Keywords:

skin cancer, melanoma, vision transformers, ensemble learning, computer-aided diagnosis

Abstract

Melanoma is the deadliest skin cancer, causing over 90% of skin tumor deaths. Early detection ensures survival above 95%, but below 15% in late stages. We propose the HEVA architecture, which integrates ensemble deep learning with ABCD descriptors and a SegFormer-B5 model, for accurate lesion masks. HEVA achieved 93.2% accuracy and 84.2% F1-score on HAM-10000, and 87.7% accuracy with 82.01% macro F1 on ISIC 2018 (1,512 images). SegFormer achieved 82.6% mean IoU and 91.2% accuracy in segmentation.

Downloads

Download data is not yet available.

References

[1] MINISTÉRIO DA SAÚDE - INCA. INCA estima 704 mil casos de câncer por ano no Brasil até 2025. 2022. In Portuguese. Disponível em: ⟨https://www.gov.br/inca/pt-br/assuntos/noticias/2022/inca-estima-704-mil-casos-de-cancer-por-ano-no-brasil-ate-2025⟩.

[2] MINISTÉRIO DA SAÚDE - INCA. Estimativa 2023 – Incidência de Câncer no Brasil. 2023. In Portuguese. Disponível em: ⟨https://www.gov.br/inca/pt-br/assuntos/cancer/numeros/estimativa⟩.

[3] ESTEVA, A. et al. Dermatologist-level classification of skin cancer with deep neural networks. Nature, v. 542, n. 7639, p. 115–118, 2017.

[4] SHORFUZZAMAN, M. An explainable stacked ensemble of deep learning models for improved melanoma skin cancer detection. Multimedia Systems, v. 28, n. 4, p. 1309–1323, 2022.

[5] BHATT, H. et al. State-of-the-art machine learning techniques for melanoma skin cancer detection and classification: a comprehensive review. Intelligent Medicine, v. 3, n. 3, p. 180–190, 2023.

[6] KITTLER, H. et al. Diagnostic accuracy of dermoscopy. The Lancet Oncology, v. 3, n. 3, p. 159–165, 2002.

[7] JAVAID, A.; SADIQ, M.; AKRAM, F. Skin cancer classification using image processing and machine learning. In: IEEE. 2021 International Bhurban Conference on Applied Sciences and Technologies (IBCAST). [S.l.], 2021. p. 439–444.

[8] VESTERGAARD, M. E. et al. Dermoscopy compared with naked eye examination for the diagnosis of primary melanoma: a meta-analysis of studies performed in a clinical setting. British Journal of Dermatology, v. 159, n. 3, p. 669–676, 2008.

[9] SHETTY, B. et al. Skin lesion classification of dermoscopic images using machine learning and convolutional neural network. Scientific Reports, v. 12, n. 1, p. 18134, 2022.

[10] BENYAHIA, S.; MEFTAH, B.; LÉZORAY, O. Multi-features extraction based on deep learning for skin lesion classification. Tissue and Cell, v. 74, p. 101701, 2022.

[11] OMEROGLU, A. N. et al. A novel soft attention-based multi-modal deep learning framework for multi-label skin lesion classification. Engineering Applications of Artificial Intelligence, v. 120, p. 105897, 2023.

[12] TSCHANDL, P.; ROSENDAHL, C.; KITTLER, H. The HAM10000 dataset: a large collection of multi-source dermatoscopic images of common pigmented skin lesions. Scientific Data, v. 5, p. 180161, 2018.

[13] SCORLETTI, F. et al. Malignant tumors misdiagnosed as benign vascular anomalies. Pediatr Blood Cancer, v. 65, n. 7, p. e27051, 2018.

[14] HEINZE, C. et al. Malignant conversion of eyelid capillary hemangioma to cutaneous angiosarcoma. Ophthalmic Plast Reconstr Surg, v. 37, n. 2, p. e54–e57, 2021.

[15] International Society for the Study of Vascular Anomalies. ISSVA Classification for Vascular Anomalies. 2018.

[16] DINIZ, J. O. B. et al. Efficientensemble: Diagnóstico de câncer de mama em imagens de ultrassom utilizando processamento de imagens e ensemble de efficientnets. In: Anais do XXIV Simpósio Brasileiro de Computação Aplicada à Saúde (SBCAS 2024). [S.l.]: SBC, 2024.