Inteligência artificial para apoio ao diagnóstico de radiografias pulmonares
Carregando...
Data
2024-07-22
Autores
Orientador
Schneiders, Luis Antônio
Banca
Pretto, Fabrício
Wolf, Alexandre Stürmer
Título do periódico
ISSN
Título do Volume
Editor
Resumo
A Inteligência Artificial (IA), consiste no uso de comportamentos e modelagens inteligentes para resolução de problemas complexos. Exames de diagnóstico por imagem, como a radiografia, tomografia computadorizada e a ultrassonografia são utilizados para o auxílio na identificação de possíveis doenças nos pacientes. Exames deste tipo geram imagens com padrões que podem ser identificados por softwares de IA, de forma a agilizar o processo de diagnóstico de doenças. Nesse sentido, o objetivo geral é apresentar uma proposta de aplicação da IA no diagnóstico de doenças pulmonares com vistas a aumentar a assertividade dos diagnósticos médicos e desenvolver uma ferramenta de IA específica para o diagnóstico de doenças pulmonares selecionadas. A pesquisa classifica-se como quali-quantitativa, de finalidade exploratória e de procedimento experimental. Para embasar a análise e desenvolvimento propostos, foi utilizado o conjunto de dados intitulado "Normal Tuberculosis Covid-19 Chest Xrays images", obtido do site Kaggle e como complemento para o desenvolvimento do trabalho, foi realizada uma solicitação formal ao Centro Clínico Univates para obtenção de imagens de exames de raios-X. A acurácia encontrada na ferramenta desenvolvida foi de 93%, o que demonstra o alto índice de precisão da IA, sendo esses promissores indicativos da aptidão do modelo na detecção e classificação das condições exigidas. Constatou-se que o modelo foi capaz de reconhecer e classificar diferentes padrões radiológicos associados a condições de saúde específicas.
Artificial Intelligence (AI) consists of the use of intelligent behaviors and models to solve complex problems. Diagnostic imaging exams, such as radiography, computed tomography and ultrasound are used to help identify possible diseases in patients. Exams of this type generate images with patterns that can be identified by AI software, to speed up the disease diagnosis process. This way, the general objective is to present a proposal for the application of AI in the diagnosis of lung diseases with a view to increasing the assertiveness of medical diagnoses and developing a specific AI tool for the diagnosis of selected lung diseases. The research was classified as qualitative-quantitative, exploratory in purpose and experimental procedure. To support the proposed analysis and development, the dataset entitled "Normal Tuberculosis Covid-19 Chest Xrays images" was used, obtained from the Kaggle website and as a complement to the development of the research, a formal request was made to the Univates Clinical Center to obtain images from X-ray exams. The accuracy found in the developed tool was 93%, which demonstrates the high accuracy rate of the AI, being indicative of the model’s ability in the detection and classification of applicable conditions. It was found that the model was able to consider and classify different radiological patterns associated with specific health conditions.
Artificial Intelligence (AI) consists of the use of intelligent behaviors and models to solve complex problems. Diagnostic imaging exams, such as radiography, computed tomography and ultrasound are used to help identify possible diseases in patients. Exams of this type generate images with patterns that can be identified by AI software, to speed up the disease diagnosis process. This way, the general objective is to present a proposal for the application of AI in the diagnosis of lung diseases with a view to increasing the assertiveness of medical diagnoses and developing a specific AI tool for the diagnosis of selected lung diseases. The research was classified as qualitative-quantitative, exploratory in purpose and experimental procedure. To support the proposed analysis and development, the dataset entitled "Normal Tuberculosis Covid-19 Chest Xrays images" was used, obtained from the Kaggle website and as a complement to the development of the research, a formal request was made to the Univates Clinical Center to obtain images from X-ray exams. The accuracy found in the developed tool was 93%, which demonstrates the high accuracy rate of the AI, being indicative of the model’s ability in the detection and classification of applicable conditions. It was found that the model was able to consider and classify different radiological patterns associated with specific health conditions.
Descrição
Palavras-chave
Inteligência Artificial; Doenças pulmonares; Diagnóstico por imagem; Artificial intelligence; Lung diseases; Imaging diagnosis
Citação
GASPERI, Celso Júnior de. INTELIGÊNCIA ARTIFICIAL PARA APOIO AO DIAGNÓSTICO DE RADIOGRAFIAS PULMONARES. 2024. Monografia (Graduação em Engenharia de Software) – Universidade do Vale do Taquari - Univates, Lajeado, 22 jul. 2024. Disponível em: http://hdl.handle.net/10737/4509.