Ultrasound imaging is one of the most widely used imaging modalities for clinical applications. Structral and functional information about the human bodies can be visualized and measured quantitatively. High-quality ultrasound images are always desired for accurate diagnosis. However, in order to obtain a refined image quality, some compromises have to be made. Firstly, for the conventional sequential systems using focused transmission, the image quality is poor. A solution is to use synthetic aperture method. Aperture synthesis technology was originated in synthetic aperture radar (SAR). The idea is to use some smaller actual apertures in order to achieve large virtual aperture which improves image quality. However, compromise has to be made between the computation complexity and the system complexity. Secondly, for the parallel systems using plane wave or diverging wave for transmission, the data acqusition rate is sufficiently high, but with conventional time-domain image reconstruction, the processing time is long for a single frame data. So the second tradeoff is between the ultrafast data acqusition and slow beamforming and inferior image quality due to the lack of transmit focus. Thirdly, because the acoustic signal is highly attenuated by the bone, the imaging across the bones cannot be realized. In this thesis, all the three challenges within ultrasound imaging have been considered and potential solutions have been proposed. All contributions are described below.

The first contribution of thesis is to study an advanced synthetic aperture beamformer which can be implemented with a very simple receive front-end. This sequential beamforming requires simple receive front-end and no delay operation has to be performed. Yet the simple first-stage beamformer significantly degrades the image quality. We aim to further improve the final image quality by looking at the second-stage dynamic beamformer. Via field pattern analysis, we propose a compounding method where two more sub-images can be reconstructed along with the original sub-image. These sub-images can be seen as being produced with different transmit origins, thus the summation of them enhances image contrast.

The second contribution of the thesis is to study new efficient Fourier beamformation methods for focused imaging. First study considers a virtual-source model. With the assumption of spherical-wave transmission at the transmit beam region, monostatic- and bistatic wavenumber algorithm can be applied. These two proposed beamformers correspond to the monostatic processing with single-line data and the bistatic processing using the subaperture data. The proposed beamformers are compared with dynamic receive focusing (DRF) and two synthetic aperture imaging algorithms. The result shows that the proposed beamformers provide better resolution for scatterers away from the focus. The image contrast is also enhanced. In addition, the computation time is significantly reduced. Depending on the pixel number, the monostatic- and bistatic beamformer shortens the computation time by up to 51 times and 2 times, resectively. Second study proposes a novel plane-wave model within the transmit beam. In this way, Fourier beamformation for plane wave imaging can be used to reconstruction sub-images in the focused imaging. Results show that the proposed time-domain plane wave model (TD-PWM) has the best CNR performance. Also, the Fourier beamformers present a more uniform resolution across the whole imaging region. Moreover, the computation time of the Fourier beamformers is shortened by 34 times compared to the conventional time-domain spherical wave model (TD-SWM).

The third contribution of the thesis is to design new Fourier beamformation for ultrafast imaging modality. For convex array, spherical wavefront can be simply synthesized by turning all elements simultaneously. Due to the lack to transmit focus, the image quality is suboptimal. One solution is to adopt virtual sources behind the transducer and compound corresponding images. We propose two novel Fourier-domain beamformers (vs1, vs2) for non-steered diverging wave imaging and an explicit interpolation scheme for virtual-source-based steered diverging wave imaging using a convex probe. The received echoes are first beamformed using the proposed beamformers and then interpolated along the range axis. The results show that the two proposed Fourier-domain beamformers give higher contrast than dynamic receive focusing (DRF) with better resolution. The computation time of vs1 and vs2, depending on the image pixel number, is shortened by 2 to 73 and 4 to 216 times than the dynamic receive focusing.

The fourth contribution of the thesis is exploiting new method for ultrasound imaging through tubid scatterers. This is made by exploiting new physics of ultrasound speckle patterns. We explore the rotational-invariant property of ultrasonic speckle and develop high-resolution speckle-scanning ultrasonography to image sub-millimeter-sized features through thick bones. We experimentally validate the rotational invariance of ultrasonic speckle. Based on this property, we scan a random ultrasonic speckle pattern across an object sandwiched between two thick bones so that the object features can be encoded to the ultrasonic waves. After receiving the transmitted ultrasonic waves, we reconstruct the image of the object using an iterative phase retrieval algorithm. We successfully demonstrate imaging of hole and tube features sized as fine as several hundreds of microns between two 0.5 to 1-cm-thick bones.