Evaluacion de la calidad de imagen en la adquisición de imágenes cerebrales utilizando el equipo PET/CT Biograph Vision 450 al variar parámetros de reconstrucción en diferentes protocolos de cerebro
Loading...
Date
Authors
Monge Cerdas, Jossue
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
La utilización del método de reconstrucción UltraHD-PET (PSF + TOF) permite realizar variaciones de los parámetros de reconstrucción de una imagen PET. Estas variaciones y sus combinaciones influyen en el ruido de la imagen final. En este trabajo, se evalúa las propiedades de la señal y ruido de la imagen en la reconstrucción de imágenes cerebrales a partir de distintos parámetros de reconstrucción utilizando el PET/CT Biograph Vision 450.
Se realiza una evaluación de un único método de reconstrucción en combinación con los siguientes parámetros: (1) Iteraciones de 1, 4, 6, 10, 20, 30, 40 (2) Filtros Gaussianos de 3mm,5mm,10mm y Filtro All Pass (3) Tiempos de adquisición de 1, 2, 3, 4, 5 min, esto en conjunto con la utilización de dos protocolos de cerebro con variaciones en la matriz de adquisición de 880 x 880 y 515 x 512 respectivamente. Se utiliza el maniquí Estriatal RSD donde el putamen y núcleo caudado izquierdo funcionan como regiones de alta captación y la cascara cerebral como región de fondo. A partir de esto se evalúa la relación señal ruido (SNR), relación contraste ruido (CNR), variabilidad de fondo(%BV) y rugosidad de imagen(%IR).
Se determina que SNR, %IR, %BV aumentan al utilizar más iteraciones y disminuyen al incrementar el tamaño del FWHM del filtro Gaussiano. La CNR disminuye con el incremento de las iteraciones y se aumenta al ampliar el tamaño del FWHM del filtro Gaussiano. Al utilizar distintos tiempos de adquisición se determina qué %BV y %IR disminuyen al utilizar mayores tiempos de adquisición y la CNR aumenta al utilizar mayores tiempos de adquisición, la SNR es independiente de este factor. Al comparar ambos protocolos se define que el protocolo B presenta una mayor SNR y el protocolo A presenta una mayor CNR, %IR y %BV, sin embargo, la diferencia de los últimos dos parámetros es baja.
The UltraHD-PET (PSF + TOF) reconstruction method allows variations in the PET image reconstruction parameters. These variations and their combinations affect the noise of the final image. In this work, image signal and noise properties are evaluated in the reconstruction of brain images from different reconstruction parameters using the PET/CT Biograph Vision 450. An evaluation of a single reconstruction method is performed in combination with the following parameters: (1) Iterations of 1,4,6,10,20,30,40 (2) Gaussian Filters of 3mm,5mm,10mm and All Pass Filter (3) Acquisition times of 1,2,3,4.5 min, this in conjunction with the use of two brain protocols with variations in the acquisition matrix of 880x880 and 515x512 respectively. The Striatal RSD phantom is used where the putamen and left caudate nucleus function as high uptake regions and the brain shell as a background region. From this, the signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), background variability (%BV) and image roughness (%IR) are evaluated. It is determined that SNR, %IR, %BV increase when using a greater number of iterations and decrease when increasing the size of the FWHM of the Gaussian filter. CNR decreases by increasing the number of iterations and increases as the size of the FWHM of the Gaussian filter increases. When using different acquisition times, it is determined that %BV and %IR decrease when using longer acquisition times and CNR increases when using longer acquisition times, SNR is independent of this factor. When comparing both protocols, it is defined that protocol B has a higher SNR and protocol A has a higher CNR, %IR and %BV, however, the difference in the last two parameters is low.
The UltraHD-PET (PSF + TOF) reconstruction method allows variations in the PET image reconstruction parameters. These variations and their combinations affect the noise of the final image. In this work, image signal and noise properties are evaluated in the reconstruction of brain images from different reconstruction parameters using the PET/CT Biograph Vision 450. An evaluation of a single reconstruction method is performed in combination with the following parameters: (1) Iterations of 1,4,6,10,20,30,40 (2) Gaussian Filters of 3mm,5mm,10mm and All Pass Filter (3) Acquisition times of 1,2,3,4.5 min, this in conjunction with the use of two brain protocols with variations in the acquisition matrix of 880x880 and 515x512 respectively. The Striatal RSD phantom is used where the putamen and left caudate nucleus function as high uptake regions and the brain shell as a background region. From this, the signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), background variability (%BV) and image roughness (%IR) are evaluated. It is determined that SNR, %IR, %BV increase when using a greater number of iterations and decrease when increasing the size of the FWHM of the Gaussian filter. CNR decreases by increasing the number of iterations and increases as the size of the FWHM of the Gaussian filter increases. When using different acquisition times, it is determined that %BV and %IR decrease when using longer acquisition times and CNR increases when using longer acquisition times, SNR is independent of this factor. When comparing both protocols, it is defined that protocol B has a higher SNR and protocol A has a higher CNR, %IR and %BV, however, the difference in the last two parameters is low.
Description
Keywords
PET/CT, Medicina Nuclear, Calidad de imagen, Imágenes cerebrales, Biograph vision 450, TOMOGRAFÍA, Tomografía por Emisión de Positrones, Reconstrucción de imagen del PET