Desenvolvimento de um software para detecção de estenose mitral a partir de ecocardiograma utilizando processamento digital de imagem : um apoio à decisão clínica

• Ano de defesa: 2025
• Autor(a) principal: Barros Filho, Genilton de França
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal da Paraíba; Brasil; Engenharia Mecânica; Programa de Pós-Graduação em Engenharia Mecânica; UFPB
• Palavras-chave em Português: Software - Ecocardiogramas - Estenose mitral; Ecocardiograma - Valva mitral; Ecocardiogramas - Impressão 3D; Processamento digital de imagens; Mitral valve; Mitral stenosis; Digital image processing; Echocardiograms; Open algorithm software; 3D printing; CNPQ::ENGENHARIAS::ENGENHARIA MECANICA
• Link de acesso: https://repositorio.ufpb.br/jspui/handle/123456789/35094

Cardiovascular Diseases (CVDs) are prevalent globally, characterized by low survival rates and an increasing incidence. Among the heart valves, the mitral valve (MV) is most frequently affected by pathological alterations, such as mitral stenosis (MS), which is defined as the incomplete opening of the valve, thereby obstructing forward blood flow. In this way, this study aims to automate the determination of the mitral valve's maximum opening area and carry out its 3D printing, using Digital Image Processing (DIP) on Three-Dimensional Transesophageal Echocardiogram (3D TEE) exams through direct planimetry, facilitated by the development of a software. The research involved thirty echocardiogram exams from thirty distinct patients, sixteen of whom exhibited normal mitral valve openings, while fourteen presented with rheumatic mitral stenosis. All videos, in AVI format, underwent DIP using Thresholding method, and a software was developed to assist in the diagnosis of stenosis and for 3D printing of mitral valves. The software successfully identified cases with and without mitral stenosis among the subjects analyzed. The maximum opening area of the valve was clearly delineated using the DIP technique. These measurements were juxtaposed with those conducted by specialists, showing remarkable similarity, with the greatest discrepancy being 0.3 cm². The correlation between these two sets of measurements was validated through the Bland-Altman method. The mitral valves were printed with FLEX and PETG material. Prints with FLEX material showed better results compared to those printed with PETG. The results of this study demonstrated the feasibility of Thresholding in detecting the presence or absence of mitral stenosis and in performing 3D printing of mitral valves from echocardiogram videos.