The effect of sound-speed-image resolution on phase aberration correction for ultrasound computed tomography

摘要

Ultrasound computed tomography (USCT) has been proposed as a hopeful tool for breast cancer screening. It can obtain three-dimensional (3D) breast images with three different modalities, automatically. The three modalities include sound-speed-image (SSI), attenuation image, and reflection image. In them, the reflection image has highest resolution, which makes it appropriate for breast cancer early detection. However, the reflection image commonly be reconstructed using a pre-determined sound speed, which has mismatch with real sound speeds in inhomogeneous breast tissues. The mismatch results in phase aberration and degrade image quality. In order to correct phase aberration, SSI can be employed for providing real sound speed information. However, SSIs reconstructed by different methods or obtained by different equipment could have different resolution. Its resolution may effect phase correction quality and reflection image quality. In this article, we evaluated the effect of SSI resolution on phase aberration correction and reflection image quality. First, USCT raw data was obtained by both numerical simulation and phantom experiment. Second, reflection images of obtained raw data were reconstructed by synthetic aperture (SA) technique, and their phase aberration was corrected by multi-stencils fast marching (MSFM) method using SSIs with different resolutions. Finally, the quality and resolution of reconstructed reflection images were evaluated. The numerical simulation quantitatively showed the diameters of points spread function (PSF) in reflection images increased from about 0.3 to 0.6 mm, while the resolution of SSI for phase correction decreasing. The phantom experiment qualitatively demonstrated both resolution and contrast of reconstructed reflection images decreased, while the resolution of SSI for phase correction decreasing. Both simulation and experiment results demonstrated that phase aberration correction performance decreases as the resolution of SSI decreasing.