Filtrado wavelet en imágenes de mamografía con contraste de fase obtenidas mediante haces de sincrotrón
Archivos
Fecha
2020-03-16
Autores
Brizuela Cardoso, Javier David
Ruiz Gonzalez, Yusely
Perez Díaz, Marlen
Título de la revista
ISSN de la revista
Título del volumen
Editor
Resumen
La innovadora técnica de mamografía con contraste de fase usando haces de sincrotrón (PCM_SR) ha mostrado cuantitativamente mejores resultados en cuanto contraste imagen y detección de pequeños detalles; pero presenta mayores niveles de ruido, respecto a las técnicas digitales convencionales hospitalarias. Este trabajo tiene cómo objetivo mostrar como utilizando el filtrado mediante la Transformada Wavelet (TW), es posible reducir el ruido a niveles semejantes a los de equipos hospitalarios, manteniendo el contraste, visibilidad de bordes y la resolución espacial. Se emplearon 23 imágenes mamográficas realizadas a un maniquí antropomórfico, obtenidas mediante PCM_SR, bajo diferentes condiciones de adquisición en cuanto a energía del haz y dosis impartida en el sincrotrón. El procesamiento se basó en un banco de filtros pertenecientes a la familia de la TW. Se cuantificó, mediante métricas, la calidad de imagen obtenida en post-filtrado resultantes como relación señal a ruido (SNR), contraste (C), visibilidad de detalles finos (V), y resolución espacial a través de la Función de Transferencia de la Modulación (MTF), sobre regiones de interés en las imágenes. Se obtuvo, que con la implementación de filtros wavelet, se observan mejoras en cuanto a calidad de imagen, al reducirse el ruido significativamente (p=0,001), sin afectar el contraste, la visibilidad y la resolución espacial. Entre los filtros del dominio de Wavelet, los filtros Sym2, y los Bior1.1 y Bior1.3, a dos niveles de descomposición, fueron los de mejores resultados superando los obtenidos por los filtros tradicionales y desempeñándose exitosamente en la reducción de ruido.
The innovative phase contrast mammography technique using synchrotron beams (PCM_SR) has shown quantitatively better results in terms of image contrast and detection of small details; but it presents higher noise levels, compared to conventional hospital digital techniques. This work aims to show how using filtering using Wavelet Transform (TW), it is possible to reduce noise to levels similar to those of hospital equipment, maintaining contrast, edge visibility and spatial resolution. 23 mammographic images performed on an anthropomorphic dummy were used, obtained by PCM_SR, under different conditions of acquisition in terms of beam energy and dose imparted in the synchrotron. The processing was based on a bank of filters belonging to the TW family. The image quality obtained in post-filtering was quantified using metrics, such as signal-to-noise ratio (SNR), contrast (C), visibility of fine details (V), and spatial resolution through the Transfer Function of the Modulation (MTF), on regions of interest in the images. It was obtained, that with the implementation of wavelet filters, improvements in image quality are observed, as the noise is significantly reduced (p = 0.001), without affecting the contrast, visibility and spatial resolution. Among the Wavelet domain filters, the Sym2 filters, and the Bior1.1 and Bior1.3, at two levels of decomposition, were the ones with the best results, surpassing those obtained by traditional filters and performing successfully in noise reduction.
The innovative phase contrast mammography technique using synchrotron beams (PCM_SR) has shown quantitatively better results in terms of image contrast and detection of small details; but it presents higher noise levels, compared to conventional hospital digital techniques. This work aims to show how using filtering using Wavelet Transform (TW), it is possible to reduce noise to levels similar to those of hospital equipment, maintaining contrast, edge visibility and spatial resolution. 23 mammographic images performed on an anthropomorphic dummy were used, obtained by PCM_SR, under different conditions of acquisition in terms of beam energy and dose imparted in the synchrotron. The processing was based on a bank of filters belonging to the TW family. The image quality obtained in post-filtering was quantified using metrics, such as signal-to-noise ratio (SNR), contrast (C), visibility of fine details (V), and spatial resolution through the Transfer Function of the Modulation (MTF), on regions of interest in the images. It was obtained, that with the implementation of wavelet filters, improvements in image quality are observed, as the noise is significantly reduced (p = 0.001), without affecting the contrast, visibility and spatial resolution. Among the Wavelet domain filters, the Sym2 filters, and the Bior1.1 and Bior1.3, at two levels of decomposition, were the ones with the best results, surpassing those obtained by traditional filters and performing successfully in noise reduction.
Descripción
Palabras clave
Mamografía con Contraste - Fase, Filtrado Wavelet, Calidad de Imagen, Contrast - Phase Mammography, Wavelet Filtering, Image Quality