Detecção de doenças pulmonares em imagens de raio x com redes convolucionais

Abstract

The detection of lung diseases using artificial intelligence has proven to be a promising approach, especially in resource-limited scenarios. The use of Convolutional Neural Networks (CNNs) has been widely explored to improve the analysis of medical images, allowing for more accurate and faster diagnoses. This study performs a comparative analysis of different approaches that apply CNNs in the detection of lung diseases, evaluating the impact of techniques such as transfer learning, image preprocessing, and hyperparameter optimization. Based on public databases, the work developed demonstrates innovative advances in the accuracy and sensitivity of the models, highlighting the potential of CNNs in improving medical diagnoses. The research reinforces the importance of artificial intelligence in healthcare, enabling more accessible diagnoses and contributing to the evolution of medical practices.