Validação de um método para análises tridimensionais de tomografia computadorizada de feixe cônico
Carregando...
Arquivos
Data
2016-12-12
Autores
Bianchi, Jonas [UNESP]
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Estadual Paulista (Unesp)
Resumo
O exame de tomografia computadorizada de feixe cônico (TCFC) tem sido utilizado amplamente na área clínica e científica médico/odontológica. Diversos softwares de diferentes fabricantes fornecem opções para processamento das imagens, segmentações e análises quantitativas tridimensionais. Um ponto que ainda gera controvérsias nessa área é a confiabilidade dos dados analisados computacionalmente devido às limitações dos algoritmos utilizados, complexidade da estrutura a ser avaliada, magnitude da mensuração, resolução espacial e variações nas metodologias de análises. Muitas dessas limitações são devidas as análises serem realizadas de formas não padronizadas e dependentes do operador. Deste modo, o objetivo geral do presente estudo foi desenvolver e validar um novo aplicativo para mensuração automática de deslocamentos ósseos a partir de TCFC de um crânio humano macerado. Para testar a confiabilidade do método, criamos um protótipo onde foram realizados deslocamentos físicos no crânio seguidos por exames de TCFC e realizamos os mesmos deslocamentos de forma virtual. As mensurações foram obtidas com base nos registros em maxila e na base do crânio por meio do nosso aplicativo e pelo 3D-Slicer, respectivamente. Além disso, realizamos uma análise visual após segmentação semiautomática por meio do ITK-SNAP para detecção do menor defeito ósseo em fragmento de osso bovino. Nossos resultados mostraram que as ferramentas testadas foram capazes de detectar deslocamentos físicos menores que a resolução espacial da imagem, sendo que os resultados foram comparáveis ao 3D-Slicer. Para os deslocamentos virtuais, foram obtidos resultados precisos, sendo que os deslocamentos foram limitados pela resolução da imagem. Além disso, observamos que a detecção e visualização de pequenos defeitos ósseos, mesmo que maiores do que a resolução espacial da imagem podem ser comprometidas pelo processo de segmentação da imagem. Concluímos que o aplicativo de análises automáticas desenvolvido é confiável para mensurações tridimensionais na área craniomaxilofacial.
Cone beam computerized tomography (CBCT) has been widely used in the clinical, scientific, medical and dental field. Various software from different manufacturers offer options for image processing, segmentation and quantitative analysis. One point that still generates controversy in this area is the reliability of the computer-analyzed data, because of the algorithms limitations, complexity of the structure to be assessed, magnitude of the measurement, spatial resolution and spatial variations in the records. Many of these limitations are caused by analyzes that are conducted in a non-standard form and highly dependent on the operator. Thus, the general objective of this study was to develop and validate a new application for automatic measurement of bone displacement from CBCTs of a macerated human skull. To evaluate our method’s reliability, we created a prototype where physical movements were performed in a human skull followed by CBCT examinations and performed the same movements virtually. Measurements were obtained from the records in jaw and skull base using our application and the 3D Slicer software, respectively. We also performed a visual analysis after the semi-automatic segmentation using the ITK-SNAP for detection of the lower bone defect in bovine bone fragment. Our results showed that we have succeeded in implementing our application of automatic analysis for three-dimensional measurements in the craniofacial area. The tools used in this study could detect physical displacements smaller than the spatial resolution of the image, and the results were comparable to the 3D Slicer software. For virtual displacements, precise results were obtained, and the movements were limited by the image resolution. Furthermore, we observed that the detection and visualization of bone defects in the bovine cortex, even higher than the spatial resolution of the image, can be compromised by the image segmentation process. We conclude that the automatic analysis derived from our developed application is reliable for three-dimensional measurements in the craniofacial area.
Cone beam computerized tomography (CBCT) has been widely used in the clinical, scientific, medical and dental field. Various software from different manufacturers offer options for image processing, segmentation and quantitative analysis. One point that still generates controversy in this area is the reliability of the computer-analyzed data, because of the algorithms limitations, complexity of the structure to be assessed, magnitude of the measurement, spatial resolution and spatial variations in the records. Many of these limitations are caused by analyzes that are conducted in a non-standard form and highly dependent on the operator. Thus, the general objective of this study was to develop and validate a new application for automatic measurement of bone displacement from CBCTs of a macerated human skull. To evaluate our method’s reliability, we created a prototype where physical movements were performed in a human skull followed by CBCT examinations and performed the same movements virtually. Measurements were obtained from the records in jaw and skull base using our application and the 3D Slicer software, respectively. We also performed a visual analysis after the semi-automatic segmentation using the ITK-SNAP for detection of the lower bone defect in bovine bone fragment. Our results showed that we have succeeded in implementing our application of automatic analysis for three-dimensional measurements in the craniofacial area. The tools used in this study could detect physical displacements smaller than the spatial resolution of the image, and the results were comparable to the 3D Slicer software. For virtual displacements, precise results were obtained, and the movements were limited by the image resolution. Furthermore, we observed that the detection and visualization of bone defects in the bovine cortex, even higher than the spatial resolution of the image, can be compromised by the image segmentation process. We conclude that the automatic analysis derived from our developed application is reliable for three-dimensional measurements in the craniofacial area.
Descrição
Palavras-chave
Tomografia computadorizada de feixe cônico, Validação de programas de computador, Computação em informática médica, Cone-beam computed tomography, Medical informatics computing, Software validation