Ultrasonic transducers enable real-time volumetric imaging
Ultrasound is a very popular imaging technology that is used in nearly all hospitals and clinics and has clear advantages over other imaging techniques, such as MRI and computed tomography. These advantages include flexibility and the absence of ionizing radiation and adverse effects.1 In recent decades, most ultrasound devices have used transducers arranged in a 1D array, which can acquire cross-sectional 2D images using piezoelectric elements (ceramics or single crystals). However, 3D ultrasound enables clinicians to make increasingly accurate diagnoses with greater understanding and confidence by enabling them to view a fetus—or pathological feature of a disease—in 3D (see Figure 1).2 To optimize the performance of 3D ultrasound, we need a transducer that generates high-quality images in real time. The most widely available type of transducer is the mechanical transducer, which is low-cost and achieves state-of-the-art image quality, although it cannot produce images in real time.
Mechanical 3D ultrasound imaging systems use linear, curved, or annular 1D array transducers, with tens or hundreds of piezoelectric elements arranged in a linear fashion. We can trigger all of these elements simultaneously or in sequence to obtain 2D images. The image slices we acquire can take different forms, depending on the method we use for signal acquisition. For example, we can use a motorized drive to move the transducer across the skin of a patient and obtain a series of 2D parallel slices, or tilt the transducer about a central or transverse axis and obtain a set of 2D images with adjustable spacing that are arranged like a fan around the axis.3
During the signal acquisition process, we store 2D images and data on the transducer position simultaneously as a volumetric data set. Combining the 2D images with information on the position of the 1D array can generate a 3D image. We use two main approaches to obtain a 3D ultrasound data set with a mechanical transducer. The first approach tracks the motion of the transducer in space, which requires an external positioning system. This method is often referred to as random or free-hand scanning. The second approach uses position sensors that are installed inside the transducer and the motion of the transducer follows a predefined profile; this is called mechanical sequential scanning. Although we have worked on increasing the speed at which the linear array moves back and forth to make the images as close to real-time as possible, we have not yet achieved the real-time performance available with a matrix array transducer.
A matrix array transducer, which generates an ultrasound pulse diverging from the array in a pyramid shape, can be considered a linear sequence of 1D phased array transducers that can form and steer an ultrasonic beam both horizontally and vertically.4 After a sound pulse is transmitted into a wide-angle cone shape by parallel processing, we receive several sound pulses, which samples the volume within the cone shape in one pass. In principle, the integrated images are true real-time 3D ultrasound images. Given the relatively small size of a 2D matrix array transducer, we consider it to be more suited to cardiac examination than to exploring the abdomen. However, recent commercial products have expanded its use to include obstetrics and gynecology. Matrix arrays have been fabricated using piezoelectric materials and by photolithography and other processing technologies from silicon wafers in the form of capacitive micromachined ultrasound transducers.5
Although we hope that matrix array transducers will ultimately replace integrated mechanical scanning transducers and other position-sensing systems, the matrix array transducer is still a developing technology. Designing a matrix array to achieve acoustic performance that is comparable to state-of-the-art 2D imaging is very difficult. The huge number of elements in the matrix array transducer prevents the use of conventional dicing and wiring technology when fabricating the acoustic module. The use of more precise material processing and fabrication or microelectromechanical systems technology is essential. In addition, the close spacing between active elements of the transducer causes extensive cross-talk between them. We have worked to minimize cross-talk between neighboring channels and increased the number of channels within a given scanning area to obtain higher-resolution images. However, there are still technical issues that must be resolved regarding electronics, including interference, achieving higher signal-processing speed, and providing the data bandwidth required for real-time volume acquisition and multiple-channel parallel signal processing.
Overall, 3D ultrasound imaging is a relatively new, exciting technology that has aroused interest in both the academic community and industry, because it provides new opportunities for generating images in patients. Ongoing developments in computing and signal processing technology now enable the acquisition, analysis, and display of volumetric imaging data almost in real time, which has facilitated many opportunities for rapid diagnosis and medical intervention. In future, 3D ultrasound imaging is expected to become a routine part of patient diagnosis and management.
Yongrae Roh received his PhD from the Pennsylvania State University, PA. After serving for four years in the Research Institute of Industrial Science and Technology, South Korea, he joined Kyungpook National University in 1994, where he is now a professor in the School of Mechanical Engineering.