Co-robotic Ultrasound Tomography: A New Paradigm for Quantitative Ultrasound Imaging
MetadataShow full item record
Ultrasound (US) is a popular medical imaging modality due to its low-cost, non-ionizing radiation, and real-time image display. However, conventional US imaging does not immediately provide quantitative information about tissue properties. US tomography is a new mode of imaging which can generate quantitative images of the object’s acoustic properties such as speed of sound (SOS). US tomographic images have shown promise in differentiating cancerous vs. normal breast tissues. The image is reconstructed based on transmission, thus requiring the US transmitters/receivers to be at different locations around the object. One recently-developed US tomography system is a circular array of US transducers used for scanning the breast. That system has limitations in terms of possible shapes and sizes of organs being scanned. Alternatively, two conventional US probes can be aligned at two sides of the organ of interest to collect tomographic data. In this work, two co-robotic setups are proposed to produce accurate and convenient alignment and localization of the two US probes. In these systems, the operator uses one probe to determine the region of interest while the other one is automatically aligned by a robotic arm. The technologies required to enable this system include calibration, robot control, application of tracking system, and force sensing. The feasibility of these setups are demonstrated in conjunction with the required components and accuracy analysis. In addition, several force sensing integration methods are demonstrated to ensure safe and smooth human-robot collaboration. The second part of this work focuses on limited-angle tomographic image reconstruction assuming two probes are aligned. We analyze the error/noise sources and their effect on reconstruction to better describe the potential and limitations of this setup. Finally, the developed technology is applied to prostate cancer imaging. The strengths and shortcomings are discussed based on simulation and phantom studies. An ex vivo prostate study is performed, and the tomography image is compared with the corresponding MRI slice as ground-truth. Future steps for translation into clinical studies are discussed. This may open new doors to low cost prostate cancer screening.