Publicação
Image Enhancement and Coronary Segmentation The First Steps Toward Three-Dimensional Reconstruction of the Coronary Arteries
| Resumo: | Cardiovascular diseases, including Coronary Artery Disease (CAD), account for a sig- nificant portion of global morbidity and mortality rates. While Coronary Angiography (CA) still remains the standard diagnostic technique for CAD, some advancements in non-invasive imaging techniques are being made. Despite the lack of non-commercial programs capable of creating a Three-Dimensional (3D) coronary model and performing real-time co-registration with non-invasive models, progress towards its development is crucial for guiding coronary intervention procedures and help planning coronary surgeries. This work aims to serve as a starting point for the development of such a program by presenting a method for selecting coronary angiographic frames and performing both image enhancement and coronary segmentation in CA and Computed Tomography Coronary Angiography (CTCA) images. In pursuit of such objectives, electrocardiographic signals were analysed, and the R peaks were detected. The CA frames chosen to be further processed were the ones that were in the same moment of the cardiac cycle as when the CTCA images were taken. For CA image enhancement, both global and local manipulations of image quality, such as brightness, contrast, and gamma adjustments, histogram equalisations, contrast- limited adaptive histogram equalisations, and edge-aware local contrast manipulations, were tested. Edge-aware local contrast manipulation was found to be the most effective for distinguishing coronary arteries from surrounding tissues. In order to segment these images, three different methods were followed: thresholding, region growing, and multi- scale vessel enhancement filtering methods, with the last one being the one with better outcomes. Finally, to enhance the CTCA images, the windowing technique was used, and then the Frangi Vesselness Filter for 3D volumes was applied, along with a simple thresholding to remove darker regions. |
|---|---|
| Autores principais: | Silva, Adriana Ponte da |
| Assunto: | Coronary Arteries Coronary Angiography Computed Tomography Coronary Angiography Image Enhancement Image Segmentation |
| Ano: | 2024 |
| País: | Portugal |
| Tipo de documento: | dissertação de mestrado |
| Tipo de acesso: | acesso aberto |
| Instituição associada: | Universidade Nova de Lisboa |
| Idioma: | inglês |
| Origem: | Repositório Institucional da UNL |
| Resumo: | Cardiovascular diseases, including Coronary Artery Disease (CAD), account for a sig- nificant portion of global morbidity and mortality rates. While Coronary Angiography (CA) still remains the standard diagnostic technique for CAD, some advancements in non-invasive imaging techniques are being made. Despite the lack of non-commercial programs capable of creating a Three-Dimensional (3D) coronary model and performing real-time co-registration with non-invasive models, progress towards its development is crucial for guiding coronary intervention procedures and help planning coronary surgeries. This work aims to serve as a starting point for the development of such a program by presenting a method for selecting coronary angiographic frames and performing both image enhancement and coronary segmentation in CA and Computed Tomography Coronary Angiography (CTCA) images. In pursuit of such objectives, electrocardiographic signals were analysed, and the R peaks were detected. The CA frames chosen to be further processed were the ones that were in the same moment of the cardiac cycle as when the CTCA images were taken. For CA image enhancement, both global and local manipulations of image quality, such as brightness, contrast, and gamma adjustments, histogram equalisations, contrast- limited adaptive histogram equalisations, and edge-aware local contrast manipulations, were tested. Edge-aware local contrast manipulation was found to be the most effective for distinguishing coronary arteries from surrounding tissues. In order to segment these images, three different methods were followed: thresholding, region growing, and multi- scale vessel enhancement filtering methods, with the last one being the one with better outcomes. Finally, to enhance the CTCA images, the windowing technique was used, and then the Frangi Vesselness Filter for 3D volumes was applied, along with a simple thresholding to remove darker regions. |
|---|