Segmentación de hematomas epidurales, usando una técnica computacional no lineal en imágenes de tomografía computarizada cerebral
No hay miniatura disponible
Fecha
2017
Autores
Vera, Miguel
Huérfano, Yoleidy
Contreras, Julio
Vera, Maria
Salazar, Williams
Vargas, Sandra
Chacón, Gerardo
Rodriguez, Jhoel
Título de la revista
ISSN de la revista
Título del volumen
Editor
Sociedad Venezolana de Farmacología Clínica y Terapéutica
Resumen
Mediante este trabajo se propone una técnica computacional
no lineal para segmentar un hematoma epidural (EDH), presente
en imágenes de tomografía computarizada multicapa.
Esta técnica consta de 4 etapas desarrolladas en el dominio
tridimensional. Ellas son: pre-procesamiento, segmentación,
pos-procesamiento y entonación de parámetros. La etapa de
pre-procesamiento se divide en dos fases. En la primera, denominada
definición de un volumen de interés (VOI), se emplea
un algoritmo de umbralización por bandas el cual permite,
fundamentalmente, acotar el EDH considerado. En la
segunda fase, identificada como filtrado, se aplica un banco
de algoritmos computacionales para disminuir el impacto de
los artefactos y atenuar el ruido presente en las imágenes. Los
algoritmos que conforman esta fase son: el filtro de erosión
morfológica (MEF) y el filtro de mediana (MF). Por otra parte,
durante la etapa de segmentación se implementa un algoritmo
de agrupamiento, denominado crecimiento de regiones (RG),
el cual es aplicado a las imágenes pre-procesadas. A fin de
compensar el efecto del MEF el EDH, segmentado preliminarmente,
es sometido a la etapa de pos-procesamiento la cual
se basa en la aplicación de un filtro de dilatación morfológica
de tipo binaria (MDF). Durante la entonación de parámetros,
el coeficiente de Dice (Dc) es utilizado para comparar las segmentaciones
dilatadas del EDH, obtenidas automáticamente,
con la segmentación del EDH generada por un neurocirujano
de manera manual. La combinación de parámetros que generan
el Dc más elevado, permite establecer los parámetros
óptimos de cada una de los algoritmos computacionales que
conforman la técnica no lineal propuesta. Los resultados obtenidos
permiten reportar un Dc superior a 0.90 lo cual indica
una buena correlación entre las segmentaciones generadas
por el experto neurocirujano y las producidas por la técnica
computacional desarrollada.
The main is to propose a non-linear computational technique to segment an epidural hematoma (EDH), present in multilayer computed tomography images. This technique consists of 4 stages developed in the three-dimensional domain: preprocessing, segmentation, post-processing and intonation of parameters. The pre-processing stage is divided into two phases. In the first one, called the definition of a volume of interest (VOI), a thresholding algorithm by bands is used, which allows, fundamentally, to delimit the EDH considered. In the second phase, identified as filtering, a bank of computational algorithms is applied to reduce the impact of the artifacts and attenuate the noise present in the images. The algorithms that make up this phase are: the morphological erosion filter (MEF) and the median filter (MF). On the other hand, during the segmentation stage a grouping algorithm is implemented, called growth of regions (RG), which is applied to the preprocessed images. In order to compensate the effect of the MEF, the EDH, preliminarily segmented, is submitted to the post-processing stage, which is based on the application of a morphological dilation filter of binary type (MDF). During the intonation of parameters, the coefficient of Dice (Dc) is used to compare the dilated segmentations of the EDH, obtained automatically, with the segmentation of the EDH generated by a neurosurgeon manually. The combination of parameters that generate the highest Dc, allows to establish the optimal parameters of each of the computational algorithms that make up the proposed nonlinear technique. The obtained results allow to report a Dc superior to 0.90 which indicates a good correlation between the segmentations generated by the expert neurosurgeon and those produced by the developed computational technique.
The main is to propose a non-linear computational technique to segment an epidural hematoma (EDH), present in multilayer computed tomography images. This technique consists of 4 stages developed in the three-dimensional domain: preprocessing, segmentation, post-processing and intonation of parameters. The pre-processing stage is divided into two phases. In the first one, called the definition of a volume of interest (VOI), a thresholding algorithm by bands is used, which allows, fundamentally, to delimit the EDH considered. In the second phase, identified as filtering, a bank of computational algorithms is applied to reduce the impact of the artifacts and attenuate the noise present in the images. The algorithms that make up this phase are: the morphological erosion filter (MEF) and the median filter (MF). On the other hand, during the segmentation stage a grouping algorithm is implemented, called growth of regions (RG), which is applied to the preprocessed images. In order to compensate the effect of the MEF, the EDH, preliminarily segmented, is submitted to the post-processing stage, which is based on the application of a morphological dilation filter of binary type (MDF). During the intonation of parameters, the coefficient of Dice (Dc) is used to compare the dilated segmentations of the EDH, obtained automatically, with the segmentation of the EDH generated by a neurosurgeon manually. The combination of parameters that generate the highest Dc, allows to establish the optimal parameters of each of the computational algorithms that make up the proposed nonlinear technique. The obtained results allow to report a Dc superior to 0.90 which indicates a good correlation between the segmentations generated by the expert neurosurgeon and those produced by the developed computational technique.
Descripción
Palabras clave
Tomografía cerebral, Hematoma epidural, Técnica computacional no lineal, Segmentación, Brain Tomography, Epidural Hematoma, Nonlinear Computational Technique, Segmentation