SISTEMA DE APOIO À DECISÃO CLÍNICA EM MAMOGRAFIA DIGITAL USANDO INTELIGÊNCIA ARTIFICIAL
SISTEMA DE APOIO À DECISÃO CLÍNICA EM MAMOGRAFIA DIGITAL USANDO INTELIGÊNCIA ARTIFICIAL
-
DOI: https://doi.org/10.22533/at.ed.3421425201110
-
Palavras-chave: Diagnóstico. Mamografia digital. Inteligência Artificial. RNC.
-
Keywords: Diagnosis. Digital Mammography. AI. CNN.
-
Abstract: This book chapter demonstrates the development and validation of a clinical decision support system for digital mammography screening using Artificial Intelligence. The proposed method integrates convolutional neural networks (CNNs) with fuzzy logic, combining the high discriminative capacity of deep learning models with the interpretability provided by rule-based inference systems. Recent studies have shown that the integration of fuzzy systems with machine learning models leads to significant improvements in diagnostic precision, particularly in clinical contexts characterized by high variability. Trained on real-world imaging data from Brazilian institutions, the system employs a convolutional architecture optimized for medical imaging tasks, processing standardized and normalized DICOM images. Each image yields an activation vector that is subsequently analyzed by a fuzzy module responsible for generating continuous and interpretable clinical inferences. The results indicate that this integrated approach enhances diagnostic sensitivity and improves model generalization, especially in cases involving incomplete data such as studies composed of only one or two images. Thus, comparison with purely neural network models reveals significant gains, positioning the solution as a promising alternative for clinical environments that require reliable and auditable tracking, with maximized accuracy and precision.
- David Lopes Maciel
- Natiele Vieira de Oliveira Maciel
- Fabrício Moraes de Almeida