UBC Theses and Dissertations
Imaging visco-elastic properties of soft tissue with ultrasound Turgay, Emre
This thesis describes a system to measure the mechanical properties of soft tissue. The mechanical properties depend on the type of tissue (e.g. fat, muscle, blood) and the presence of disease or pathology. For example, it has been known that many cancers, such as carcinoma of the breast and the prostate, appear as hard nodules. Manual palpation is widely used for detecting these nodules but detection by palpation is restricted to large tumors that reside relatively close to an accessible surface. Current imaging devices such as computed tomography, magnetic resonance imaging and ultrasound are not directly capable of measuring the mechanical properties of tissue. An emerging research topic, called elastography, aims to produce a new type of image that depicts mechanical properties of tissue. The basic principle is to excite motion in a tissue, record the motion with an imaging device, and then estimate the mechanical properties from the recorded data. This thesis presents a new approach based on this principle where a computer-controlled vibrator induces motion over a range of frequencies simultaneously and the resulting displacement is recorded at multiple locations and time instants with a sequence of ultrasound images. Two methods are proposed for estimating mechanical properties such as stiffness, damping, and mass, from the recorded data. In the first method, the tissue is modeled by using mass elements at different locations connected to each other by springs and dampers. These elements represent the local mass, stiffness, and damping of the tissue. The equation of motion for the proposed one dimensional model is solved to extract these parameters. The second method performs a transfer function analysis of the tissue motion. According to this method, the tissue dynamics between two locations along the axis of motion is considered as a linear dynamic system. The transfer function between the two locations is obtained by spectral analysis with the recorded motion used as inputs and outputs. The shape of the transfer function can then be analyzed further. For example, the stiffness of tissue can be estimated from the magnitudes of the transfer function at low-frequencies. Common to both methods is a new time domain correlation-based tracking algorithm for measuring tissue motion in successive ultrasound images. The algorithm is based on stretching the time domain ultrasound signals according to the local compression applied along the axial direction. The accuracy of the new method is demonstrated to be higher than the standard time domain correlation-based algorithms on a small number of tests on tissue mimicking materials. Both simulations and experiments have been performed to validate the proposed methods. Simulations show that the parameters can be extracted within 1 % error when the measurement noise level is set to 0 %, for one dimensional mass-spring-damper models with varying parameters. The simulations are repeated by taking ultrasonic noise and correlation-based motion tracking noise into account. The errors in damping and stiffness values are within 5 % at 4 mm resolution, and 12 % at 6 mm resolution for mass values, for a simulated one dimensional homogeneous tissue. Initial experimental results from homogeneous and layered tissue mimics are reported which demonstrate the ability of both methods to quantitatively image tissue stiffness. Preliminary results on tissue damping distribution for homogeneous tissue mimics are found to be comparable to the reported damping values for the same type of material in the literature. The shear viscosity values are found to be between 12.5-19.7 Pa-s for a gelatin block with a Young's modulus of 25 kPa . Extraction of mass is still under investigation. Both the simulations and the experimental results show that the new methods are feasible and further investigation is needed to continue development of the ideas.
Item Citations and Data