Welcome to the Duke University
Professor Stephen Smith
Deformable array transducers for phase aberration correction in medical ultrasonic imaging
Phase aberrations due to inhomogeneities in the acoustic velocity of human tissues degrade medical ultrasound images by disrupting the ultrasound beam focus. Near field phase correction algorithms compensate for aberrating tissue located close to the transducer by adjusting the electronic phase delays used to steer and focus the ultrasound beam. In order to correct the two-dimensional phase aberrations in tissue using this technique, a two-dimensional array is necessary. However, two-dimensional arrays are a complex option for phase correction due to their large number of elements and poor sensitivity. Instead of using a full two-dimensional array, a new technique is proposed which uses a deformable transducer of significantly fewer channels for two-dimensional phase correction. Phase correction in azimuth is achieved by altering the electronic phase delay of the elements. However, phase correction in elevation is achieved by tilting the elements in elevation with a piezoelectric actuator. Simulations of phase aberration correction using a deformable array transducer were compared to electronic correction with a 2-D array. The results have shown that a deformable 1 x N or 2 x N array transducer can approach the image quality of a 4 x N two-dimensional array. A prototype 1 x 32 deformable array was developed using a RAINBOW actuator for both the element deflection and the generation of ultrasound. The prototype array was characterized with measurements of vector impedance, pulse-echo sensitivity, and bandwidth. Phase correction in elevation was simulated by tilting the elements on-line to alter the B-scan image. Measured cyst contrast of a tissue mimicking phantom increased from 0.76 with half the elements tilted to 0.86 for the corrected case with all the elements aligned. To improve the future performance of the deformable array while minimizing the fabrication effort, 2-D and 3-D finite element analysis (FEA) was developed to predict the characteristics of the deformable array. Since the deformable array combines a mechanical actuator with a medical ultrasound transducer, both the low frequency acutator and ultrasonic characteristics of the array were modeled and were well matched to experimental results.
Design and fabrication of multiplexed two-dimensional transducer arrays using electrostrictive ceramic materials
Two-dimensional array transducers are essential for real time volumetric imaging. The implementation of receive mode parallel processing requires the transducer be designed with a wide transmit beam to allow multiple simultaneous receive beams. Unfortunately, this requirement has an associated increase in pulse-echo beamwidth that reduces contrast of subtle lesions. When the volumetric system is used for B-mode scanning, the wide transmit beam degrades image quality unnecessarily. It is hypothesized that a multiplexed two-dimensional array transducer can be developed to improve image quality of a volumetric scanner when it is used for B-mode scanning, while maintaining the capability of real time volumetric imaging. It is further hypothesized that 2-D array multiplexing can be achieved using the bias controlled sensitivity of electrostrictive relaxor ferroelectric materials. The hypothesis was tested by analysis, development and evaluation of sparse random arrays and relaxor ferroelectric materials. Random array patterns with Gaussian and uniform element distributions were analyzed by computer simulation. A random array pattern was designed for each mode of the multiplexed transducer. Beamwidth was controlled by distribution of the 192 transmitter and 64 receivers, whereas the average sidelobe amplitude was determined by the number of elements. Bar mode piezoelectric properties of two prototype relaxor ferroelectric materials were measured and used in KLM transducer modeling. The materials exhibited high dielectric permittivity, but low electromechanical coupling and high dielectric loss. Bias circuits were designed to allow operation of the relaxor transducer with the existing phased array system. The 3.5 MHz, 82 x 82 multiplexed transducer was fabricated with 428 active elements. Element and array performance were in good agreement with predictions. Multiplexing was successfully accomplished with performance comparable to commercial multiplexors. The multiplexing system was integrated with the Duke phased array scanner and allowed rapid multiplexing between element patterns. Images of test phantoms indicated the element pattern for B-mode imaging had improved contrast when compared to the element pattern for volumetric imaging.
Signal-to-noise ratio of transducer arrays for medical ultrasound
Linear array transducers for medical ultrasound would be valuable for field use and emergency room treatment. Such a portable ultrasouund scanner requires reducing the size and weight of the system and minimizing the amount of power necessary for scanning. For example, integrating the transmit and receive circuitry by decreasing the conventional transmit voltage from 100 V to 15 V reduces the size of the scanner. Unfortunately, the transmit sensitivity is reduced because of the reduction in transmit voltage. In receive mode, the linear array element combined with the loading effects of the coaxial cable limit the received signal-to-noise ratio (SNR). Therefore, the pulse-echo SNR of a linear array has to be optimized to make imaging feasible. It was hypothesized the pulse-echo SNR could be improved using a hybrid array. In transmit, the poor sensitivity in a portable ultrasound scanner and a 2-D array was improved by using multilayer lead zirconate titanate (PZT) ceramic elements which reduced the element input impedance and increased the power coupled into the body. In addition to decreasing the element impedance, a reduction in the transmitter source impedance from the typical value of 50 $/Omega$ to 7 $/Omega$ further increased transmit sensitivity. The received SNR was improved by using a single layer PZT element combined with a high impedance preamplifier adjacent to the element. The low impedance transmitters were also located in the handle with the receive preamplifiers. The hypothesis was tested by fabricating and testing a 5 MHz hybrid linear phased array. The 5 MHz hybrid array which consisted of 64 elements had alternating multilayer transmit and single layer receive elements. A low voltage transmitter with an output impedance of 7 $/Omega$ and a high impedance preamplifier were placed adjacent to the transmit and receive elements. B-scan images of tissue phantoms as well as abdominal images using the Siemens SI-1200 scanner further confirmed the hypothesis. To further improve the ultrasound scanner maneuverability, an optoelectronic transmitter was investigated to replace the conventional electronic transmitter. Flexible light weight fiber optics replaced the coaxial cable. A photosensor in the handle switched wide bandwidth pulses across the transducer element upon detection of an optical signal. Feasibility studies were performed using a 500 mW AlGaAs laser diode and a silicon photoconductive switch (PCSS). Finally, a 48 channel ultrasound system with 16 optoelectronic transmitters and 32 conventional electronic receivers was designed to interface with a Siemens SI-1200 scanner and a 2.25 MHz linear array. Transmit signal measurements and B-scan images of cysts and tumors in tissue mimicking phantoms were performed to compare the optoelectronic transmitter to the conventional electronic transmitter. (Abstract shortened by UMI.)
Real-time 3D ultrasound guidance of interventional devices
Using real-time 3D (RT3D) ultrasound during minimally invasive procedures can improve interventional device positioning. This work describes the integration of a RT3D ultrasound system with four techniques that may enhance interventional device guidance. These methods include: (1) vibrating devices and visualizing their guidance using RT3D color Doppler, (2) using multiple transducers to improve guidance and therapy delivery, (3) displaying stereo pairs for real-time stereoscopic 3D ultrasound, and (4) combining robotic catheter systems with real-time 3D ultrasound. The feasibility of using a RT3D ultrasound system was tested in tissue phantoms and in animal models using four interventional devices coupled to a vibrating piezoelectric element. An analytical model was developed to approximate the vibrating motion of the interventional devices. The application of this model combined with modifications to the RT3D ultrasound system increased the color Doppler magnitude signals by 14 dB, which improved the detection of two clinically relevant interventional devices during an in vivo dog study. The development of scanning with multiple 3D ultrasound catheters was investigated. The goal was to use one catheter transducer survey cardiac anatomy and to use the second transducer to home in on the target tissue. Once into position, the second transducer was used to monitor therapy delivery. Preliminary imaging studies showed the system was successful in tissue phantoms and an in vivo dog study. A third method to enhance guidance involved displaying stereo image pairs in real-time. The stereo image pairs were merged into a single image and provided an added sensation of depth. This system was tested by viewing previously-saved volumes of ultrasound data in stereo 3D, including volumes of interventional devices. A feasibility study investigated using RT3D ultrasound images from catheter transducers to automatically guide a robotic arm during a simulated interventional procedure. Accuracy tests performed in a water tank indicate that the catheter transducers can be used to guide devices using a robot with a 10% error. Experiments using a vascular graft suspended in a water tank showed an ability to guide a needle toward a bright target within the graft. The feasibility of acoustic radiation force impulse (ARFI) imaging using a 2D matrix array and a RT3D ultrasound system was also investigated.
Integrated catheters for three-dimensional intracardiac echocardiography and ultrasound ablation
Two recent advances have expanded the role of ultrasound in intracardiac medicine: the introduction of real-time 3-D intracardiac echocardiography and the development of intracardiac ultrasound ablation systems. This work describes catheters combining both technologies. Such devices could potentially supplant the radiofrequency catheters guided by fluoroscopy that are used now to treat cardiac arrhythmias like atrial fibrillation. A series of devices are designed, built, and tested. The first combines a 64 element 2-D array for imaging at 5MHz and a 2 mm by 4 mm piston ablation transducer operating at 10 MHz. The second device is a forward-scanning catheter integrating a 112 element 2-D array for imaging at 5 MHz encircled by an ablation annulus operating at 10 MHz. Both devices are capable of imaging heart anatomy like the atria, ventricles, and septum, produce spatial peak, temporal average intensity ( ISPTA ) of 30.0 W/cm2 and 16.1 W/cm2 respectively, and can ablate ex vivo bovine tissue in less than 2 minutes. The temperature rise and lesion formation from the ablation annulus are modeled using acoustic field simulations and finite element analysis. These results match experimental temperature rise within 2°C and lesion size within 1 mm in diameter and depth. The model is then used to simulate linear phased array transducers for ablation. Two linear phased array transducers are built to confirm the finite element analysis findings, one with 86 elements operating at 8 MHz and one with 50 elements operating at 4 MHz. The 4 MHz array shows promise for ablation but the real-time 3-D ultrasound scanner is incapable of driving the array. A mechanically focused 86 element 8 MHz array is described. This device transmits ISPTA of 29.3 W/cm2 and creates a lesion in 2 minutes.
Multi-layer piezoelectric ceramics for medical ultrasound transducers
In medical ultrasound imaging, 2-D array transducers have become essential to implement dynamic focusing and phase-correction in the elevation dimension as well as real-time volumetric scanning. Unfortunately, the small size of a 2-D array element results in a small clamped capacitance and a large electrical impedance near resonance. These elements have poor signal-to-noise ratio (SNR) because their impedance is much higher than the electrical impedance of the transmit and receive circuitry. It is hypothesized that the SNR can be improved by using an N layer structure of piezoelectric ceramic with the layers connected acoustically in series and electrically in parallel. Further, it is hypothesized that such multi-layer ceramics (MLC's) can be accurately modeled by a simplified circuit model, the KLM transmission line model, and finite element analysis. For the MLC, the clamped capacitance is multiplied by a factor of N$/sp2$ and the electrical impedance by 1/N$/sp2$ compared to a single layer element of the same dimensions. The hypothesis was tested by analysis, development and performance measurement of MLC array transducers. Prototype 2-D array elements were fabricated from 19 layer PZT-5A, operating at 1 MHz. The electrical impedance was 50 $/Omega,$ compared to 14 k$/Omega$ for a single layer control element. As a result, the pulse-echo SNR increased by 26.3 dB for the MLC compared to the control when driving a coaxial cable load on receive. Also, a 2.25 MHz, 3 x 43 phased-array transducer was fabricated using 3 layer PZT-5H material. The MLC was manufactured using thick film technology with plated-through vias to electrically interconnect the electrode layers. The MLC elements had an impedance of 115 $/Omega,$ compared to 830 $/Omega$ for single layer control elements. For the low impedance MLC elements, the pulse-echo SNR increased by 6.8 dB when driving a coaxial cable load. B-scan images were made of cysts in a tissue-mimicking phantom and of the left kidney in vivo. The images clearly showed a higher signal-to-noise ratio for the MLC array compared to the control. As a result, 2-D arrays made of multi-layer PZT can be used to form images at a higher frequency and a greater penetration depth than single layer arrays. These results were consistent with simulations from the KLM model and finite element analysis. For future applications of MLC's, further improvement in SNR can be obtained using a hybrid design for a sparsely sampled array. Several hybrid array designs were analyzed.
Ultrasound catheter transducers for intracranial brain imaging and therapy
Each year, over 13,000 people in the United States die from a primary malignant brain tumor. Currently, primary BTs are treated most commonly by surgery, radiotherapy, and systemic chemotherapy, though each of these methods carries a risk of complications or acute side effects. Ultrasound hyperthermia has been investigated as way to open the blood-brain barrier for improved chemotherapeutic drug delivery, but previous methods have involved either invasively removing skull bone via surgery or non-invasively dealing with the high ultrasound attenuation, reflection, and phase aberration resulting from the skull and its variable thickness. Dual-mode ultrasound transducers for image-guided therapy have also been investigated for several applications; in some instances, phased arrays are ideal, allowing control over the ultrasound energy deposition pattern and inherent spatial registration between imaging, treatment, and monitoring. Additionally, thermosensitive liposomes can be configured to encapsulate drugs and actively target regions of tumor angiogenesis. When used in combination with localized hyperthermia, thermosensitive liposomes can provide targeted control of drug release that may enhance chemotherapeutic efficacy in many clinical settings. Meanwhile, catheter devices and endovascular techniques are used by interventional neuroradiologists to treat various intracranial diseases, including intracranial aneurysm and dural venous sinus thrombosis. These procedures can be extended to the treatment of intracranial tumors (advancement of a 5 Fr catheter as far as the frontal portion of the superior sagittal sinus has been demonstrated). The objective of the work presented in this dissertation was the realization of a dual-mode catheter transducer for a minimally-invasive, vascular approach to deliver localized, image-guided ultrasound hyperthermia to an intracranial tumor target. Toward this end, a series of prototype ultrasound transducers were designed, simulated, built, and tested for imaging and therapeutic potential. Two 14-Fr phased-array prototypes were built with PZT-5H ceramic and tested for real-time 3D intracranial imaging and focused-beam hyperthermia capability. These were able to visualize the lateral ventricles and Circle of Willis in a canine model, and generate a temperature rise over 4°C at a 2-cm focal distance in excised tissue. Single-channel intravascular ultrasound (IVUS) coronary imaging catheters as small as 3.5 Fr were then considered as a construction template; several possible transducer apertures were simulated before fabricating prototypes with PZT-4. The transducers exhibited a dual-frequency response, due to the presence of thickness-mode and width-mode resonances. A thermal model was developed to estimate the +4°C thermal penetration depth for a given transducer aperture, predicting an effective therapeutic range of up to 12 mm with a 5 × 0.5 mm aperture. A 3.5-Fr commercial mechanical IVUS catheter was retrofitted with a PZT-4 transducer and tested for 9-MHz imaging performance in several animal studies, successfully visualizing anatomical structures in the brain and navigating a minimally-invasive vascular pathway toward the brain. An identical PZT-4 transducer was used to build a 3.3-MHz therapy prototype, which produced a temperature rise of +13.5°C at a depth of 1.5 mm in live xenograft brain tumor tissue in the mouse model. These studies indicate that a minimally-invasive catheter transducer can be made capable of visualizing brain structures and generating localized hyperthermia to trigger drug release from thermosensitive liposomes in brain tumor tissue. gated.
Phase aberration correction for real-time 3D transcranial ultrasound imaging
Phase correction has the potential to increase the image quality of real-time 3D (RT3D) ultrasound, especially for transcranial ultrasound. Such improvement would increase the diagnostic utility of transcranial ultrasound, leading to improvements in stroke diagnosis, treatment, and monitoring. This work describes the implementation of the multi-lag least-squares cross-correlation and partial array speckle brightness methods for static and moving targets and the investigation of contrast-enhanced (CE) RT3D transcranial ultrasound. The feasibility of using phase aberration correction with 2D arrays and RT3D ultrasound was investigated. Using the multi-lag cross-correlation method on electronic and physical aberrators, we showed the ability of 3D phase aberration correction to increase anechoic cyst identification, image brightness, contrast-to-noise ratio (CNR), and, in 3D color Doppler experiments, the ability to visualize flow. With a physical aberrator, CNR increased by 13%, while the number of detectable cysts increased from 4.3 to 7.7. We performed an institutional review board (IRB) approved clinical trial to assess the ability of a novel ultrasound technique, namely RT3D CE transcranial ultrasound. Using micro-bubble contrast agent, we scanned 17 healthy volunteers via a single temporal window and 9 via the sub-occipital window and report our detection rates for the major cerebral vessels. In 82% of subjects, we identified the ipsilateral circle of Willis from the temporal window, and in 65% we imaged the entire circle of Willis. From the sub-occipital window, we detected the entire vertebrobasilar circulation in 22% of subjects, and in 50% the basilar artery. We then compared the performance of the multi-lag cross-correlation method with partial array reference on static and moving targets for an electronic aberrator. After showing that the multi-lag method performs better, we evaluated its performance with a physical aberrator. Using static targets, the correction resulted in an average contrast increase of 22.2%, compared to 13.2% using moving targets. The CNR increased by 20.5% and 12.8%, respectively. Doppler signal strength and number of Doppler voxels increased, by 5.6% and 14.4%, respectively, for the static method, and 9.3% and 4.9% for moving targets. We performed two successful in vivo aberration corrections. We used this data and measure the isoplanatic patch size to be an average of 10.1 degrees. The number of Doppler voxels increased by 38.6% and 19.2% for the two corrections. In both volunteers, correction enabled the visualization of a vessel not present in the uncorrected volume. These results are promising, and could potentially have a significant impact on public health.
Real-time three-dimensional intracardiac ultrasound imaging using two-dimensional catheter arrays
Intracardiac echocardiography (ICE) is a minimally invasive imaging technique in which a miniaturized ultrasound transducer is mounted in the tip of a catheter, enabling image acquisition from within the heart. To date, many applications of ICE involve the guidance of cardiac interventional procedures, such as ablation treatment of atrial fibrillation, guidance of atrial septal puncture. Current commercially available ICE systems offer monoplanar imaging, i.e. imaging in a 2-D plane, acquired using a mechanically rotating single transducer element, or with a linear phased array of elements. With these 2-D imaging configurations, it is often difficult to orient the imaging catheter in such a way that both the cardiac anatomy and interventional device are aligned in the same imaging plane. The development of a real-time three-dimensional (RT3-D) ICE system addresses these shortfalls. This work describes the design, simulation, fabrication and testing of miniaturized, 2-D phased array transducers mounted in the tips of catheters to enable RT3-D ICE. The transducers are constructed on multi-layer polyimide interconnect circuits and incorporate high density cabling interconnections within the catheter lumen. Advances in interconnection and cabling technology are described which have enabled up to a 5X increase in channel density over previous designs. These advances have facilitated both the miniaturization of the devices and image quality improvements necessary for the devices to become clinically useful. Several designs are discussed, including a 9 Fr (3.0 mm O.D.), 70 element 7 MHz side-viewing 2-D catheter array on a silicon substrate, and a 7 Fr (2.3 mm O.D.), 112 element 5 MHz side-viewing 2-D catheter array. RT3-D images of cardiac anatomy are presented, obtained during in vitro and in vivo studies with a sheep model. The design, fabrication and testing of forward-viewing RT3-D ICE probes is also described. The forward-viewing probes contain an additional working lumen through which interventional devices are delivered. The forward viewing probes contain 112 elements operating at 5 MHz. Combining the imaging catheter with a working lumen in a single device may simplify cardiac interventional procedures by allowing clinicians to easily visualize cardiac structures and simultaneously direct interventional tools in a RT3-D image.
Two-Dimensional Arrays for Real Time Volumetric Imaging
The design, fabrication, and evaluation of two dimensional array transducers for real time volumetric imaging are described. The transducers we have previously described operated at frequencies below 3 MHz and were unwieldy to the operator because of the interconnect schemes used in connecting to the transducer handle. Several new transducers have been developed using new connection technology. A 40 x 40 = 1600 element 3.5 MHz array with –6 dB fractional bandwidth of 63%, 50 Ohm insertion loss of –63 dB, and a –6 dB pulse-echo angular response of 35° was fabricated with 256 transmit and 256 receive elements. A 60 x 60 = 3600 element 5.0 MHz array with –6 dB fractional bandwidth of 50%, a 50 Ohm insertion loss of -69 dB, and a –6 dB pulse-echo angular response of 18° was constructed with 248 transmit and 256 receive elements. An 80 x 80 = 6400 element 2.5 MHz array with a –6 dB fractional bandwidth of 54%, a 50 Ohm insertion loss of -64 dB, and a –6 dB pulse-echo angular response of 14° was fabricated with 256 transmit and 208 receive elements. An 11 x 13 = 143 element 5.0 MHz array with a –6 dB fractional bandwidth of 50%, a 50 Ohm insertion loss of -64 dB , and a –6 dB pulse-echo angular response of 36° for use in an intracardiac catheter was constructed with 51 transmit and 30 receive elements. All the transducers were used to generate real time volumetric images in phantoms and in vivo using the Duke University real time volumetric imaging system which is capable of generating multiple planes at any desired angle and depth within the pyramidal volume.
The Ultrasound brain helmet: Simultaneous multi-transducer 3D transcranial ultrasound imaging
My work examines the problem of rapid imaging of stroke and presents ultrasound-based approaches for addressing it. Specifically, the aberration and attenuation due to the skull are discussed as sources of image degradation and a prototype system for simultaneous 3D bilateral imaging via both temporal acoustic windows is proposed. This system uses custom sparse array transducers built on flexible multilayer circuits that can be positioned for simultaneous imaging via both temporal acoustic windows, allowing for registration and fusion of multiple real-time 3D scans of cerebral vasculature. I examine hardware considerations for new matrix arrays—-transducer design and interconnects—-for this application. Specifically, it is proposed that signal-to-noise ratio (SNR) may be increased by reducing the length of probe cables. This claim is evaluated as part of the presented system through simulation, experimental data, and in vivo imaging. Ultimately, gains in SNR of 7 dB are realized by replacing a standard probe cable with a much shorter flex interconnect; higher gains may be possible using ribbon-based probe cables. In vivo images are presented depicting cerebral arteries with and without the use of microbubble contrast agent that have been registered and fused using a search algorithm which maximizes normalized cross-correlation. The scanning geometry of a brain helmet-type system is also utilized to allow each matrix array to serve as a correction source for the opposing array. Aberration is estimated using cross-correlation of RF channel signals followed by least mean squares solution of the resulting overdetermined system. Delay maps are updated and real-time 3D scanning resumes. A first attempt is made at using multiple arrival time maps to correct multiple unique aberrators within a single transcranial imaging volume, i.e. several isoplanatic patches. This adaptive imaging technique, which uses steered unfocused waves transmitted by the opposing or “beacon” array, updates the transmit and receive delays of 5 isoplanatic patches within a 64°×64° volume. In phantom experiments, color flow voxels above a common threshold have increased by an average of 92% while color flow variance decreased by an average of 10%. This approach has been applied to both temporal acoustic windows of two human subjects, yielding increases in echo brightness in 5 isoplanatic patches with a mean value of 24.3 ± 9.1%, suggesting such a technique may be beneficial in the future for improving image quality in non-invasive 3D color flow imaging of cerebrovascular disease including stroke. Acoustic window failure and the possibility of overcoming it using a low frequency, large aperture array are also examined. In performing transcranial ultrasound examinations, 8-29% of patients in a general population may present with window failure, in which it is not possible to acquire clinically useful sonographic information through the temporal acoustic window. The incidence of window failure is higher in the elderly and in populations of African descent, making window failure an important concern for stroke imaging through the intact skull. To this end, I describe the technical considerations, design, and fabrication of low-frequency (1.2 MHz), large aperture (25.3 mm) sparse matrix array transducers for 3D imaging in the event of window failure. These transducers are integrated into the existing system for real-time 3D bilateral transcranial imaging and color flow imaging capabilities at 1.2 MHz are directly compared with arrays operating at 1.8 MHz in a flow phantom with approximately 47 dB/cm0.8/MHz0.8 attenuators. In vivo contrast-enhanced imaging allowed visualization of the arteries of the Circle of Willis in 5 of 5 subjects and 8 of 10 sides of the head despite probe placement outside of the acoustic window. Results suggest that the decrease from approximately 2 to 1 MHz for 3D transcranial ultrasound may be sufficient to allow acquisition of useful images either in individuals with poor windows or outside of the temporal acoustic window by untrained operators in the field.
Multi-layer composite transducer arrays for improved signal-to-noise ratio and bandwidth in medical ultrasound
Increasing transducer bandwidth and signal-to-noise ratio (SNR) is fundamental to improving the quality of medical ultrasound images. In this dissertation, I describe a range of array transducers using new materials to improve both parameters. These new materials are stacked multi-layer composites of piezoelectric ceramics and polymer epoxies. The first transducer consists of 2 layers of posts (piezoelectric ceramic, PZT-5H) surrounded by soft epoxy. Experimentally, this 2 layer 1-3 composite transducer, yielded increased pulse-echo SNR by 5.2 dB and increased -6 dB bandwidth by a factor of 1.3, compared to the PZT-5H control. However, this structure required precision alignment of the posts greater than 90% of the post pitch (0.125mm) and a thin film bond line between the layers. Thus, I developed a new multi-layer structure that will not require post alignment and would ideally be fabricated using thick film technology capable of volume production. Starting from a PZT-5H multi-layer transducer, cuts were made through the top layer and back-filled with epoxy, forming a PZT/epoxy composite layer on top of PZT layers, referred to as a multi-layer composite hybrid transducer. Finite element simulations (FEM) showed that for a 2 MHz phased array element with a single acoustic matching layer, the 3 layer hybrid structure increases the pulse-echo SNR by 11 dB compared to a single layer PZT-5H control element and increases -6 dB pulse-echo fractional bandwidth from 46% to 65%, a factor of 1.4, for the hybrid element. I fabricated a hybrid transducer array and obtained improvement in SNR by 11 dB over a PZT-5H control and increased -6 dB bandwidth from 54% to 59%, a factor of 1.1. However, the material properties of currently available thick film multi-layer transducers limit the performance of these hybrid arrays and need further refinement before simulated results can be matched experimentally. Additional FEM simulations were performed to further improve the transducer array designs. These simulations showed that for a 2 MHz phased array element with a single matching layer, the improved 3 layer hybrid structure increased the pulse-echo SNR by 16 dB and -6 dB pulse-echo fractional bandwidth from 58% to 75%, a factor of 1.3, for the hybrid element versus the PZT-5H control. Analogous FEM simulations of single crystal material (PZN-8%PT), showed increased pulse-echo SNR by only 3.1 dB versus the PZT-5H control and a -6 dB bandwidth of 108%.
Real-time 3D ultrasound endoscopy
The development of real-time three-dimensional (RT3D) ultrasound imaging for endoscopic procedures can improve diagnostic imaging as well as serve as a guidance and assessment tool for surgeries and other interventional purposes. This work describes the development and fabrication of an endoscope-based device that is capable of RT3D ultrasound imaging and the testing of this device as an instrument for diagnosis, guidance, and therapy in a variety of clinical applications. The first steerable endoscopic RT3D probe was constructed for the Volumetrics Medical Imaging (VMI) scanner. The device is 1cm in diameter and was fabricated on a multilayer flexible interconnect circuit designed for 504 elements in a 6.3mm x 6.3mm aperture. Incorporation of high density ribbon cabling enabled the high channel count of the transducer. This device, operating at 5MHz, was tested to have an average -6dB bandwidth of 25.3%. The completed probe was used to image in vivo canine anatomy, specifically the heart, esophageal wall, liver, and prostate. Real-time 3-D transesophageal scans provided clear visualization of the heart valves, coronary sinus, and pulmonary veins. Panoramic 120° scans with the endoscopic device demonstrated the utility of the probe for interventional guidance in electrophysiological procedures. For laparoscopy, the probe was also used successfully in post-mortem canine studies integrating coordinates acquired from the ultrasound scans with a robot linear motion system to direct a needle to targets in abdominal organs. In vitro and post-mortem experiments with the robot demonstrated an average guidance error of less than 2mm. The probe was also used to determine the feasibility of combined endoscopic RT3D ultrasound and hyperthermia. Acoustic field simulations and finite element analysis of the probe have shown that intensities of over 3.43 W/cm 2 are required to induce temperature rises in tissue of 4°C at a 3cm focus. With modifications to the VMI scanner, the probe produced 2.40 W/cm2 using 8 cycle bursts at a pulse repetition frequency of 8 kHz, yielding a maximum ex vivo temperature elevation 2.3°C. In addition, in order to increase the field of view close to the transducer face, the first cylindrical curvilinear matrix array for RT3D ultrasound was developed.
Real-time rectilinear 3-D ultrasound imaging
Current real-time volumetric scanners use a 2-D array to scan a pyramidal volume consisting of many sector scans stacked in the elevation direction. This scan format is primarily useful for cardiac imaging to avoid interference from the ribs. However, a real-time rectilinear volumetric scanner with a wider field of view close to the transducer could prove more useful for abdominal, breast, or vascular imaging. This work describes the design and development of the first real-time rectilinear 3-D ultrasound scanner. The system featured three sparse 2-D array designs producing increasing image quality. The first array was a 5 MHz Mills cross array. It consisted of a 2 x 94 transmit arm and a perpendicular 94 x 2 receive arm. The Duke prototype 3-D ultrasound scanner, T4, was modified for real-time rectilinear volumetric imaging by changing the beamformer and display software. The field of view was 30 x 8 x 60 mm. The second transducer was a 5 MHz 2-D periodic array having 169 transmitters and 256 receivers. The receivers measured 0.3 x 0.3 mm and were spaced every 2.4 mm in azimuth and elevation. The transmitters had dimensions of 2.4 x 2.4 mm and a pitch of 2.4 mm. The Model 1 3-D scanner built by Volumetrics Medical Imaging (Durham, NC) was modified for real-time rectilinear volumetric scanning by changing the beamformer and display software. This array had an increased field of view of 30 x 30 x 60 mm. To improve the imaging quality of the periodic array, a new 5 MHz array was built having 1024 receivers and used 4:1 receive mode multiplexers. The receivers were 0.6 x 0.6 mm and had a staggered distribution to suppress grating lobes. Simulations indicated a 13 dB increase in pulse-echo sensitivity and another 13 dB decrease in grating lobe levels compared to the periodic array. Images of tissue phantoms and in vivo showed significant improvement in penetration and contrast compared to the periodic array. Images from a 1.5 cm diameter cyst in a phantom showed a 12 dB improvement in sensitivity and a 6 dB improvement in contrast.
Ultrasound and optical scanners using micromachine technology
Numerous applications benefit from both high frequency ultrasound imaging and optical imaging techniques. High frequency ultrasound imaging is used in intracardiac and intravascular imaging as well as imaging the skin, eye, and small animals for genetic studies. Potential uses of optical scanners range from bar code scanners and laser printers in industry to corneal resurfacing and optical coherence tomography in medicine. A major issue in developing these systems is steering the acoustic and optical beams. This work describes the design, fabrication and testing of new types of ultrasound and optical beam scanning devices. These devices are fabricated from polyimide films using photolithography and use a linear polyimide MEMS actuator to mechanically scan the beams. This actuator, the integrated force array (IFA), is a network of hundreds of thousands of micron scale deformable capacitors that electrostatically contract with an applied voltage. Forward viewing tables pivoting on cantilever hinges and side scanning structures tilting on torsion hinges were fabricated on polyimide substrates with tables 1.125 mm and 2.25 mm wide. These structures were modeled using one dimensional beam theory and ANSYS finite element analysis prior to fabrication. For the ultrasound probes, PZT transducers were fabricated on these tables that operate at 20 MHz and 30 MHz and yielded insertion losses of 20-26 dB with fractional bandwidths of 34-49%. The transducer assemblies driven by MEMS actuators produced sector scans of 20-45° in air at resonant frequencies of 32 Hz to 90 Hz and sector scans in fluid of 6-8°. Both forward viewing and side scanning devices were then used in conjunction with a single channel high frequency ultrasound system to make real time images of wire phantoms. The optical scanning devices had gold-coated silicon mirrors mounted on the table in place of the PZT transducers. Only side scanning devices were developed for optical applications. For environmental protection, the devices were conformally coated with 500 angstroms of parylene. These devices demonstrated optical scan angles up to 146° for applied voltages up to ±50 volts. These devices were also used to steer a laser beam in a prototype bar code scanner to demonstrate functionality.