Aplicação de pós-filtros adaptativos aos beamformers mínima variância e eigenspace para formação de imagens de ultrassom utilizando a técnica de ondas planas

Abstract

Recently, new adaptive beamformers methods using the plane wave technique have been proposed by several authors in order to improve the quality of medical images formed by ultrasound. Many of these methods, although promising, are relatively new and have room for further studies and development. The main objective of this work was to combine the Minimum Variance Generalized Sidelobe Canceller (MVGSC) and Eigenspace Generalized Sidelobe Canceller (EBGSC) beamformers with the adaptive post-filters Frost, Modified Frost, Gamma Map, Lee, Kuan and their updated shapes to improve the quality of ultrasound images using flat. All simulated data for the development of this work were obtained from the Plane Wave Challenge in Medical Ultrasound Imaging website. The results were compared with the Delayand-Sum method, as it is the most widespread, and with the beamformers themselves without the use of filters. The MVGSC-Frost and EBGSC-Frost methods stood out in relation to contrast, having a better performance than the other methods, but generating less accurate images. The modified MVGSC-Frost and modified EBGSC-Frost performed similarly in relation to contrast when compared to the MVGSC and EBGSC methods, with these methods standing out in relation to contrast/noise and resolution. The Enhanced Frost post-filter combined with beamformer techniques generated unwanted noise in the images, proving to be unpromising. The MVGSC-Gamma MAP, MVGSC-Kuan, Enhanced MVGSC-Kuan, MVGSC-Lee, Enhanced MVGSC-Lee, EBGSC-Gamma MAP, EBGSC-Kuan, Enhanced EBGSC-Kuan, Enhanced EBGSC-Lee for EBGSC-Lee methods the formation of ultrasound images by decreasing the noise in the images. Some of the methods in this work helped to improve medical ultrasound images, with emphasis on the techniques MVGSC-Frost, EBGSCFrost, Modified MVGSC-Frost and Modified EBGSC-Frost that generated images with high contrast using less computational resources.