A DEEP-LEARNING BASED TOOL FOR HISTOPATHOLOGICAL DIAGNOSIS OF MELANOMA USING CONVOLUTIONAL NEURAL NETWORKS

Autores

DOI:

https://doi.org/10.16891/2317-434X.v13.e4.a2026.id2848

Palavras-chave:

Inteligência Artificial, Patologia, Melanoma

Resumo

The application of deep learning to medical image analysis to aid diagnosis in healthcare has emerged as a powerful tool for early disease detection, thus improving treatment prospects and patient recovery. The aim of this study was to develop a convolutional neural network (CNN)-based predictive model to support the histopathological diagnosis of melanoma. To develop the proposed model, a dataset consisting of 411 images was used, including 393 of these in the experimental phase. The dataset was divided into 70% of the images for training and 30% for testing, and a model was constructed using ResNet50 architecture. The results showed that ResNet50 rapidly acquired the ability to distinguish features to accurately perform histopathological melanoma diagnoses. The error rate rapidly converged, achieving accuracy of approximately 90%. This proposed model is able to enhance diagnostic accuracy and support clinical practice in melanoma detection.

Biografia do Autor

Thiago Magalhães Amaral, Federal University of Vale do São Francisco – UNIVASF

Biomedical Engineer, holding a Master’s and Ph.D. in Production Engineering with a focus on Operational Research from the Federal University of Pernambuco. Specialist in Hospital Management from Hospital Sírio Libanês and in Data Science from the Federal University of Technology – Paraná. He worked for over five years as Head of Planning at Ebserh in the University Hospital of the Federal University of Vale do São Francisco (HU-Univasf). He was a professor in the Master’s Program in Public Administration at Univasf and served as the Director of Institutional Development at Univasf. He coordinated the Health Innovation Consortium in Petrolina-PE (a project funded by FACEPE) and was a consultant for the United Nations Educational, Scientific and Cultural Organization (UNESCO) at the Municipal Department of Education of São Paulo. He is currently an Associate Professor at the Federal University of Vale do São Francisco in the Production Engineering program, a member of the Innovation Chamber at FACEPE, and a professor in the Graduate Program in Intellectual Property and Technology Transfer for Innovation (PROFNIT-Univasf).

Jefferson Tales Oliva, Federal University of Technology - Paraná – UTFPR

Bachelor in Computer Science and Master in Dynamic and Energy Systems Engineering from the Western Paraná State University (UNIOESTE), Foz do Iguaçu campus. He earned his Ph.D. from the Graduate Program in Computer Science and Computational Mathematics at the Institute of Mathematical and Computer Sciences (ICMC) of the University of São Paulo (USP), São Carlos campus. He is currently an Assistant Professor in the Computer Engineering program at the Federal University of Technology – Paraná (UTFPR), Pato Branco campus. In 2017, part of his doctoral research was conducted at Massachusetts General Hospital/Harvard Medical School. He is also a collaborating researcher at the Bioinformatics Laboratory (LABI) at UNIOESTE/Foz do Iguaçu. His expertise lies in Computer Science and multidisciplinary applications, with research interests in data mining, machine learning, signal processing, natural language processing, image analysis, biomedical informatics, and AutoML.

Henrique Takashi Idogava, Federal University of Vale do São Francisco – UNIVASF

Holds a Bachelor’s and Master’s degree in Mechanical Engineering from the State University of Campinas (UNICAMP), with co-supervision from the Center for Three-Dimensional Technologies at Renato Archer Information Technology Center (CTI – Campinas). He earned his Ph.D. from the School of Engineering of São Carlos at the University of São Paulo (EESC-USP), with research focused on Additive Manufacturing. He participated in the Print (Institutional Internationalization Program) with a Capes Sandwich Ph.D. scholarship for research at the Faculty of Engineering of the University of Porto (FEUP-Portugal). He is currently a professor in the Mechanical Engineering Department at the Federal University of Vale do São Francisco (CENMEC – UNIVASF), Juazeiro campus (BA), and a professor in the Graduate Program in Materials Science (CPGCM-UNIVASF). In the field of extension, he coordinates the Cactus Rockets Design Model Rocketry Extension Project and the INOVA Network Project. He also coordinates the Interdisciplinary Additive Manufacturing Laboratory (NIMA – UNIVASF), registered on the National Research Infrastructure Platform of the Ministry of Science, Technology, and Innovation (MCTI).

Downloads

Publicado

2026-01-24

Como Citar

Amaral, T. M., Oliva, J. T., & Idogava, H. T. (2026). A DEEP-LEARNING BASED TOOL FOR HISTOPATHOLOGICAL DIAGNOSIS OF MELANOMA USING CONVOLUTIONAL NEURAL NETWORKS. Revista Interfaces: Saúde, Humanas E Tecnologia, 13(4), 6233–6245. https://doi.org/10.16891/2317-434X.v13.e4.a2026.id2848

Edição

v. 13 n. 4 (2025):

Seção

Artigos

Artigos mais lidos pelo mesmo(s) autor(es)