Development and implementation of a shear wave speed estimation algorithm for crawling waves sonoelastography
No Thumbnail Available
Date
2024-10-10
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Pontificia Universidad Católica del Perú
Abstract
La elastografía por ultrasonido es una técnica de imagen no invasiva cuyo objetivo es brindar información acerca de la elasticidad de los tejidos biológicos evaluando sus propiedades biomecánicas. Proporciona información valiosa sobre su rigidez, la cual está relacionada a los cambios biomecánicos ocasionados por condiciones patológicas. Esto la convierte en una herramienta valiosa para el diagnóstico y el seguimiento del tratamiento de enfermedades como el cáncer. Uno de los métodos de elastografía cuantitativa se basa en el uso de dos fuentes de vibración para generar un patrón de interferencia en el tejido. La onda de corte puede visualizarse en tiempo real mediante sonoelastografía y debe aplicarse un esquema de inversión para recuperar la velocidad de la onda de corte a partir de varios fotogramas. Sin embargo, los estimadores estudiados en la literatura exhiben algunas limitaciones tales como un pobre rendimiento en entornos ruidosos y prolongados tiempos de adquisición y procesamiento para aplicaciones en tiempo real. En esta tesis se proponen e implementan dos algoritmos, basados en la transformada de Fourier de tiempo corto y la transformada Wavelet continua, para la estimación de la velocidad de la onda de corte. Se fabricaron maniquíes homogéneos y heterogéneos de gelatina para ´ evaluar el rendimiento de los algoritmos en condiciones controladas. Se realizaron experimentos de sonoelastografía a diferentes frecuencias de vibración para evaluar su precisión en una serie de escenarios. Los resultados demuestran que los algoritmos desarrollados son comparables a los estimadores existentes en términos de sesgo, coeficiente de variación, relación contraste-ruido y resolución. En medios homogéneos, el coeficiente de variación se mantuvo debajo del 10% para ambos estimadores y, en medios heterogéneos, la transformada Wavelet continua alcanzo una relación contraste-ruido de 30 dB en promedio. En general, los algoritmos muestran una robustez superior, sobre todo en presencia de una relación señal-ruido deficiente y en tejidos más rígidos con velocidades de onda de corte más elevadas. Se demostró que los algoritmos propuestos no necesitan todo el vídeo de sonoelastografía para generar un mapa de velocidad de la onda de corte, sino un solo fotograma. Esto permite la visualización en tiempo real de la velocidad de la onda de corte, lo que puede beneficiar a diversas aplicaciones clínicas.
Ultrasound elastography is a noninvasive imaging technique that aims to provide information about the elasticity of biological tissues by evaluating their biomechanical properties. It provides valuable information about their stiffness, which is related to biomechanical changes caused by pathological conditions. This makes it a valuable tool for diagnosing and monitoring the treatment of diseases such as cancer. One of the quantitative elastography methods is based on the use of two vibration sources to generate an interference pattern in the tissue. The shear wave can be visualized in real time by sonoelastography and an inversion scheme must be applied to recover the shear wave velocity from several frames. However, estimators studied in the literature exhibit some limitations such as poor performance in noisy environments, and long acquisition and processing times for real-time applications. In this thesis, two algorithms, based on the Short-Time Fourier Transform and the Continuous Wavelet Transform, are proposed and implemented for shear wave velocity estimation. Homogeneous and heterogeneous gelatin phantoms were fabricated to evaluate the performance of the algorithms under controlled conditions. Sonoelastography experiments were performed at different vibration frequencies to evaluate their accuracy in a range of scenarios. The results show that the developed algorithms are comparable to existing estimators in terms of bias, coefficient of variation, contrast-to-noise ratio and resolution. In homogeneous media, the coefficient of variation remained below 10% for both estimators and, in heterogeneous media, the Continuous Wavelet Transform achieved a contrast-to-noise ratio of 30 dB on average. In general, the algorithms show superior robustness, especially in the presence of poor signal-to-noise ratio and in stiffer tissues with higher shear wave velocities. It was shown that the proposed algorithms do not need the entire sonoelastography video to generate a shear wave velocity map, but only a single frame. This allows real-time visualization of the shear wave velocity, which can benefit various clinical applications.
Ultrasound elastography is a noninvasive imaging technique that aims to provide information about the elasticity of biological tissues by evaluating their biomechanical properties. It provides valuable information about their stiffness, which is related to biomechanical changes caused by pathological conditions. This makes it a valuable tool for diagnosing and monitoring the treatment of diseases such as cancer. One of the quantitative elastography methods is based on the use of two vibration sources to generate an interference pattern in the tissue. The shear wave can be visualized in real time by sonoelastography and an inversion scheme must be applied to recover the shear wave velocity from several frames. However, estimators studied in the literature exhibit some limitations such as poor performance in noisy environments, and long acquisition and processing times for real-time applications. In this thesis, two algorithms, based on the Short-Time Fourier Transform and the Continuous Wavelet Transform, are proposed and implemented for shear wave velocity estimation. Homogeneous and heterogeneous gelatin phantoms were fabricated to evaluate the performance of the algorithms under controlled conditions. Sonoelastography experiments were performed at different vibration frequencies to evaluate their accuracy in a range of scenarios. The results show that the developed algorithms are comparable to existing estimators in terms of bias, coefficient of variation, contrast-to-noise ratio and resolution. In homogeneous media, the coefficient of variation remained below 10% for both estimators and, in heterogeneous media, the Continuous Wavelet Transform achieved a contrast-to-noise ratio of 30 dB on average. In general, the algorithms show superior robustness, especially in the presence of poor signal-to-noise ratio and in stiffer tissues with higher shear wave velocities. It was shown that the proposed algorithms do not need the entire sonoelastography video to generate a shear wave velocity map, but only a single frame. This allows real-time visualization of the shear wave velocity, which can benefit various clinical applications.
Description
Keywords
Elastografía, Imagen ultrasónica, Ultrasonido--Biomedicina
Citation
Collections
Endorsement
Review
Supplemented By
Referenced By
Creative Commons license
Except where otherwised noted, this item's license is described as info:eu-repo/semantics/openAccess