Ultrasound Imaging: Infrastructure on Improved Imaging Methods

Report of the OWH/NCI Sponsored Workshop: March 26^th 1999

Overview

The above workshop topic addressed the potential development of research interfaces for commercially available clinical ultrasonic scanners as a basis for improved infrastructure for ultrasound research. Workshop attendees included representatives from medical industry, clinical and academic research communities, and federal agencies (NCI, OWH, NSF, and FDA). A very strong consensus was reached among the workshop participants that the development of these interfaces would markedly advance: (a) the development of new clinical applications of ultrasound, (b) ultrasonic scanner and transducer technology, and (c) the utility of ultrasonic imaging in small animal pre-clinical models and other basic biological research applications. The industry representatives, in particular, agreed that the proposed interface was realistic and could be developed in a timely manner. Recommendations included the infrastructure support necessary for the development of scanner interfaces that will involve industry and academic partnerships with industry. Recommendations included a request that NCI should proceed with reviewing novel mechanisms for supporting this type of research infrastructure that would have a wide impact on the field of research in ultrasound imaging.

This report is organized as outlined below:

1. Prior Workshop Agenda: March 26, 1999

2. Workshop Recommendations
2.1 Infrastructure Recommendations
2.2 Specific Recommendations: Ultrasonic Scanner Research Interfaces

Appendix A. Rationale for Development of Ultrasonic Scanner Research Interface
A1: Clinical Applications
A2: Basic Science and pre-clinical applications

Appendix B: Proposed Specifications for Ultrasonic Scanner Research Interface
B1: Overview
B2: General Interface Characteristics
B3: Basic Research Interface
B4: Advanced Research Interface

Appendix C: Workshop Attendees

1. Prior Workshop Agenda: March 26, 1999

On March 26, 1999, the U.S. Public Health Service’s Office on Women’s Health (OWH) and the National Cancer Institute (NCI) sponsored a workshop entitled, "Ultrasound Imaging: Infrastructure for Improved Imaging Methods". The proposed agenda of the workshop included:

• To identify the scope of expanded ultrasound clinical driven investigations if improved access to the scanner controls was facilitated by industry, with emphasis on woman’s health issues.
• To determine the feasibility of expanding current limited industry-academic partnerships to allow: (a) enhanced access to the ultrasound system for improved scanner programming and raw data acquisition, and (b) a mechanism for the use of experimental probes.
• To explore the feasibility of using "MRI models for RF pulse sequence programming interface" that: (a) allows for a user friendly development of imaging methods, that incorporates parameter limits for FDA regulations on exposure and safety issues, and (b) limited use of experimental probes with incorporated control parameters.
• To explore: (a) the short term feasibility of prioritizing a subset of programming tools that would be of immediate benefit to basic and clinical research investigators, and (b) establish a framework for distribution of such software packages, (c) explore mechanisms for further expansion of such methods, and (d) to formulate recommendations re NCI’s potential role.
• To investigate: (a) the feasibility of establishing industry/academia partnerships with the objective of both the development of software tools and organizing user groups to improve technology transfer, (i.e. improve research time lines for both senior and young investigators in industry and academia), (b) review means to address proprietary and license (patent) issues in relation to controlled access to such software methods, (c) discuss improved training strategies for basic and clinical ultrasound researchers specifically within the context of expanded access to these specialized research tools, and (d) to formulate recommendations re: NCI’s potential role.

2. Workshop Recommendations

2.1 Infrastructure recommendations

Participants in the workshop included high-level representatives from each of four ultrasonic scanner manufacturers, two clinical and five academic medical ultrasound researchers, and government representatives from the NCI, OWH, NSF, and FDA. The participants reached a broad and surprisingly strong consensus on each of the key points considered at the workshop. The major conclusions reached by the panel are:

• The development of ultrasonic scanner and transducer technology, of new clinical applications of ultrasound, of ultrasound’s role in small animal imaging and other basic biology applications, of new imaging modes such as contrast-assisted imaging and harmonic imaging, and of combined modality and surgery-assist imaging instruments, are all significantly hampered by the inability of clinical and academic researchers to access the controls of commercial ultrasonic scanners. Most NIH funded ultrasonic research is significantly hampered by the absence of such control interfaces.
• The development of research interfaces to current and future models of ultrasonic scanners is readily possible. These interfaces would allow a wide range of control over pulse sequences, pulse intensity, focusing, steering and other parameters. The acquisition of raw, quantitative echo data is an essential feature of these interfaces.
• The development of research interfaces to commercial ultrasonic scanners can only be achieved with the active participation and considerable efforts of the scanner manufacturers. The evaluation and future evolution of the interfaces will require formal collaborations between the manufacturers and academic and clinical researchers.
• The research interfaces will necessarily be quite sophisticated and their successful utilization will require talented users and extended training periods. The development of interface related user groups, academic-university run training programs, and detailed training manuals are essential for their widespread use and long-term success.

2.2 Specific Recommendations: Ultrasonic Scanner Research Interfaces

The obvious ideal scenario would be the creation of extremely powerful research interfaces (i.e. those that allow extensive control over a wide range of system parameters and storage of large amounts of raw data) by a large number of manufacturers. Widespread and long-term utilization will require the creation of training manuals and courses, the formation of user groups, and the recreation of these research interfaces on subsequent generations of ultrasonic scanners.

The workshop participants agreed that the development and continued availability of ultrasonic scanner research interfaces could only be achieved with the strong support and considerable efforts of the scanner manufacturers. These interfaces can only function as software add-ons to continuously evolving scanner control software and their creation necessarily relies on access to proprietary information.

The workshop participants also agreed that participation by academic and clinical researchers, the expected major users of the research interfaces, is essential in defining the features of the interfaces and in evaluating their working properties.

The workshop attendees make the following recommendations to the NCI:

1. That the NCI seek methods to fund the creation of research interfaces to ultrasonic scanners.
2. That the research interfaces include most or all of the proposed features included in the Appendix to this report.
3. That the NCI explore obtaining the financial participation of other NIH Institutes, the NSF, the Whitaker Foundation, and other funding agencies in supporting these interfaces.
4. That funding be directed towards partnerships between ultrasonic scanner manufacturers and academic/research institutions to interactively create and evaluate the research interfaces and to develop the accompanying user manuals. In addition, these partnerships should foster the creation of user groups and training courses for the interfaces.

The technical specifications of the proposed interface are described in Appendix B.

Appendix A. Rationale for Development of Ultrasonic Scanner Research Interface

A1: Clinical Applications

Broad Scope of Clinical Applications: It was the consensus of the workshop that the development of an ultrasound research interface would have a tremendous impact on ultrasound imaging and therefore on medical imaging as a whole. Ultrasound procedures currently comprise a significant fraction of non-film based imaging. Using the University of Virginia as an example, approximately 20,000 ultrasound and 20,000 computed tomography studies are performed each year, in comparison with 9,000 magnetic resonance studies and 10,000 nuclear medicine studies. Thus, ultrasound imaging represents approximately one third of these procedures. Ultrasound procedures also comprise a significant fraction of imaging reimbursements. Medicare reimbursements for echocardiography ($740 million in 1993) alone are more than twice that of all MR procedures. Thus, ultrasound is emerging as a leading imaging modality in terms of its clinical impact. However, the number of researchers with access to research instruments is very small, a fraction of those with research access to magnetic resonance imaging systems. Research interfaces to ultrasound systems are currently available only to those in the device manufacturing industry and very few selected researchers. Many investigators funded by major grants spend a large fraction of their efforts attempting to duplicate the resources that would be immediately available with a research interface to a commercial system.

Impact of Technical Advances for Imaging: A wide range of new basic science and clinical applications for ultrasound imaging are on the horizon. While just a few years ago ultrasound was considered to be a mature technology, new applications and technical improvements indicate that this is clearly not the case. As a result of improvements in transducers and beam-formation, a higher operating frequency and therefore significantly improved spatial resolution, is now a reality. Current systems have operating frequencies up to 15 MHz, and thus an axial spatial resolution on the order of 100 microns, where only a few years ago the maximum center frequency was 7 MHz. This improved spatial resolution provides the opportunity to resolve structures such as discrete layers of the carotid wall. Other technical improvements have included digital acquisition sub-systems and multi-dimensional arrays, further improving resolution as well as dynamic range, and vastly increasing the research flexibility of the systems. The ability to focus a beam in each plane greatly improves image quality, as well as providing the opportunity to quantify echogenicity at all depths of interest.

Interventional Procedures: Much of the recent increase in the number of ultrasound procedures has been due to its expanded use in interventional procedures. Ultrasound is now often used to guide breast biopsy, to monitor cardiac surgery, and to guide angioplasty. While this has been a significant change, the use of ultrasound to guide surgery is expected to continue its rapid expansion, due both to the benefit to the patient and the decrease in the cost of the intervention. The RDOG study has found that core needle biopsy costs only 30 to 40% as much as surgical biopsy. The use of ultrasound to guide biopsy is an example of an area in which further improvement could have a substantial impact. An improvement in image quality, providing the opportunity to find microcalcifications, or the development of contrast agents that allow the visualization of angiogenic lesions in small tumors could allow ultrasound-guided biopsy to further replace stereotactic X-ray guidance and surgical biopsy. Providing a research interface to enable the integration of the interventional and imaging modes could greatly speed this development.

Harmonic Imaging: Harmonic or nonlinear imaging has revolutionized ultrasound just in the past two years. Several recent clinical studies in the breast, abdomen, thyroid and heart show that the large majority of patients demonstrate improved image quality with harmonic imaging over conventional imaging. Optimization of imaging techniques, array technologies, and signal processing strategies associated with harmonic imaging is only beginning. Models of the mechanisms of image quality improvement by harmonic imaging and of harmonic signal generation in tissue are still quite crude. Experimental studies of harmonic imaging require the control of the intensity, timing, and polarity of ultrasonic pulses, as described in the proposed research interface specifications.

Multi-Modality Imaging: Another research area that would benefit from the development of this interface is the combined application of multiple imaging modalities. In order to develop techniques to register images from MR, CT, nuclear medicine and ultrasound, a uniform spatial grid must be created. This requires knowledge of the transmission parameters of the system (beamwidth, etc). Also, non-linear image processing applied to each imaging modality must be removed in order to directly compare and combine the resulting data. Three-dimensional reconstruction is also a growing area of research interest in all imaging modalities. While the principal interest in ultrasound has been in obstetrics, the use of 3D ultrasound in cardiology and peripheral vascular disease is also of great interest. In order to evaluate 3D interpolation and image processing techniques, the raw (beam-formed) data is required.

Ultrasound contrast agents: Contrast agents show the potential to greatly extend the utility of ultrasound imaging while significantly increasing the sophistication. Echoes produced from these agents are sufficiently strong that an individual microbubble can be detected by typical ultrasound instrumentation. An ultrasound pulse sequence is a set of pulses delivered to a single line-of-sight with a change in the pulse amplitude, phase, and/or interpulse timing within the sequence. Recent results indicate that the echo from a bubble may be uniquely recognized by its response to a pulse sequence that alternates the phase or amplitude of transmission. The resulting images from these techniques may allow the differentiation of tumors from normal tissue, and the identification of ischemic cardiac tissues. A research interface is required to create flexible transmitted pulse sequences, and to store the echoes from each transmitted pulse. With the storage of the raw received echoes, multiple signal processing schemes can be evaluated. While current contrast-assisted imaging techniques have concentrated on estimating vascular volume, contrast-assisted ultrasound also has a unique opportunity to estimate micro-vascular flow rate. In order to estimate microvascular flow rate, a train of pulses with an increasing inter-pulse interval is transmitted. The first pulse in each case can destroy all microbubbles within the small volume. The amplitude of the echo from the second pulse is then used to determine the volume of contrast agents that has re-entered the region. As the inter-pulse time is increased, the microvascular flow rate is estimated by the slope associated with the restoration of the signal amplitude. Since the region of interrogation is on the order of hundreds of microns, this seems to be a very promising technique to locally evaluate tissue perfusion, an opportunity that is unique to ultrasound.

Elastography: Elastography provides the opportunity to evaluate the mechanical properties of tissue using ultrasound as a probe. In clinical applications that include the detection of tumors and kidney disease, increased or decreased tissue stiffness can be detected. In the optimization of these measurements, the transmitted beam parameters must be optimized to recognize small lesions. Also detection algorithms must be embedded within the ultrasound system in order to evaluate performance on a real-time basis. Post-processing with delayed presentation of an image is inadequate to evaluate potential algorithms in a clinical setting.

A2: Basic Science and Pre-Clinical Applications

Ultrasound will clearly play a role in the basic scientific evaluation of genetically engineered animal models. It has been shown that single element transducers operating at a frequency of 20 MHz can successfully estimate blood velocity in the mouse heart and peripheral vascular system. Commercial systems now provide transducers with a center frequency of 15 MHz, with a 20 MHz center frequency to be available soon. This provides adequate spatial resolution to examine many structural and functional properties of small animal models. Thus, data acquisition with commercial ultrasound systems can now fulfill the role of special purpose experimental systems in basic scientific investigations. In addition, a high-resolution ultrasound scan can be obtained in a few seconds and therefore far less instrumentation is required to support the physiologic functions of the animal, in comparison with magnetic resonance imaging. Thus, ultrasound has begun to play a role in the high-speed screening of animal models and provides the opportunity to efficiently detect defects in genetically engineered animal models and embryos. The estimation of blood flow in discrete vessels and the microvasculature is also a particularly powerful capability of ultrasound in the functional evaluation of these animal models. In order for commercial systems to be applied to these small animal models, optimization of system parameters for mouse imaging must be placed under the control of the researcher. Current commercial systems are not optimized for the small vessels, rapid heart rate, and increased shear rate relevant to the mouse model. The proposed research interface would be extremely useful in this development. It has also been shown that the real-time capability of ultrasound can be useful in developmental biology. Under ultrasound guidance, labeled cells can be injected into the developing embryo, and the eventual cell migration monitored with optical imaging or magnetic resonance imaging. It is the very high frame rate (>100 Hz) and the ease of use that uniquely identify ultrasound as the modality of choice to monitor such real-time interventions. Again, the system parameters must be optimized for this very special case of a restricted field of view, near field target and required high frame rate.

In summary, ultrasound imaging instrumentation is progressing rapidly, providing wide-ranging new opportunities. In order to capitalize on this potential, clinical, engineering and basic science researchers require access to transmission parameters and received echoes. A small investment in the development of this interface could produce enormous gains in research productivity.

Appendix B: Proposed Specifications for Ultrasonic Scanner Research Interface

B1: Overview

The workshop participants propose two levels of research interface specifications: Basic and Advanced. The Basic Research Interface is expected to meet the needs of the large majority of researchers and be achievable on the current generation of scanners. The Advanced Research Interface provides additional user-controlled scanner parameters would be significantly more challenging to create but would allow more powerful experimental designs.

Both interface levels are expected to require only software development by the scanner manufacturers (i.e. no new boards).

B2: General Interface Characteristics

The interface, at either level, is intended to allow the user to:

1. Find targets of interest using conventional B-mode and Doppler methods.
2. Enter into a "research mode" with user-programmed scanner parameters.
3. Store raw echo signals during operation in the research mode.

"Raw signals", in this context, are defined as digital data, either radio-frequency (rf) signals or the real and imaginary parts of the complex envelope (quadrature signals). Recorded waveforms should be beamformed echo signals resulting from a single transmit focus that have not been subjected to nonlinear processing, e.g., logarithmic amplitude compression. Waveforms must be recorded after the application of time-gain compensation (TGC). For quadrature data, a single demodulation frequency must apply to the entire duration of the echo line recorded. All of the echo data displayed in the image should be recorded in a data file. A region of interest is selected by zooming the image. There must be facilities for externally triggered digital data acquisition. Standard apodization functions should be applied, e.g., rectangular and Hanning. The user interface must execute commands through a single system prompt. The display should indicate the min and max A/D values to show the user how much of the dynamic range is being used. The time between any request for digital data and the acquisition must be known. It should be kept to a minimum, e.g., less than one second. Although no standardized data file format is specified, there must exist a file header with all the information necessary to fully interpret recorded data. It must be possible to read the file, including header information, with Matlab or other standard technical computing languages. Information about digitization rates, filtering, external triggering, TGC settings, calibrated output power and overall gain settings, frame-recording time stamp, and all aspects of beamforming must be included.

Safety: The system display must clearly indicate to any user that a research mode is active. This must include a prominent message on the display(s) and an exterior label warning that a nonclinical mode is possible. The system must automatically revert back to normal clinical mode when the patient’s name is changed.

B3: Basic Research Interface

A) B-Mode Specifications

It must be possible to record digital data during real-time imaging for most B-mode settings. Memory should be sufficient to record digital data for .5 seconds at conventional frame rates.

The user is provided limited control of the receive beamformer. Included are the ability to select aperture size, focal point, field-of-view, aperture position, and the ability to disable aperture growth and dynamic focusing.

The frame rate for digital acquisition must be programmable. The highest rate should approach conventionally utilized rates, however, a programmable lower frame rate must be available.

B) Color Flow, M-Mode, and Pulsed Doppler Specifications

The user is given limited control of packet size and interleave ratio. Capture of all digital data for a specified region of interest is provided. For PW and M-Mode, the complete echo line is captured (no range gate). Header information must include all the intrinsic system parameters necessary to reconstruct echo signals, including firing sequence and frame rate. The minimum consecutive dwell time for a digital data acquisition must be a record length of .5 seconds.

B4: Advanced Research Interface

Implementation was proposed in three stages. These specifications are in addition to those for the basic RI.

C) Level 1

The user is provided increased user control over receive beamforming. Included are the ability to specify apodization, line density, prf, and beam steering. Receive properties associated with transmit frequency and pulse length must be adjustable.

Digital data must be available from individual receive channels.

D) Level 2

Provide control over transmit pulse length, frequency, polarity of the excitation pulse, amplitude, and pulse sequencing. FDA power limits may be ignored by assuming the user has obtained IRB approval for human use. The manufacturer will identify modes that satisfy FDA limits and those that do not.

E) Level 3

Provide separate control over pulsed transmission on individual elements.

Appendix C: Workshop Attendees.

NCI Representatives

Laurence (Larry) Clarke PhD., Workshop Organizer: Point of Contact.
Chief: Office of Imaging Technology
Diagnostic Imaging Program
Division of Cancer Treatment and Diagnosis
National Cancer Institute, EPN 800
6130 Executive Boulevard, MSC 7400
Bethesda, MD 20892-7440
Phone: 301 496 9362
Fax: 301 480 5785
Email: LC148m@nih.gov

Daniel C. Sullivan MD
Associate Director, Diagnostic Imaging Program, NCI
EPN, Room 800
6130 Executive Blvd.
Rockville, MD 20892-7440
Tel: 301 496 9531
Fax: 301 480 5785
sullivand@mail.nci.nih.gov or ds274k@nih.gov

Attendees:

Gregg Trahey, Ph.D. Chair of Workshop
Department of Biomedical Engineering,
Box 90281 Campus
Duke University
Durham, NC 27708
Phone: 919 660 5169
Email: get@egr.duke.edu

Kathy Farrara, Ph.D. ( Moving to new address late spring)
Department of Biomedical Engineering
Box 377
Health Science Center
University of Virginia
Charlottesville, Virginia 22908
Express mail:
1105 West Main Street’
Room 1421, Charlottesville, Virginia 22908
Phone: 804 243 6321
Email: Kathyf@virginia.edu

J. Brian Fowlkes, Ph.D.
University of Michigan Health Systems
Department of Radiology
Kresge III, R3315
200 Zina Pitcher Place
Ann Arbor, MI 48109-0553
Phone: (734) 763-5882
Fax: (734) 764-8541
Email: fowlkes@umich.edu

Barry B. Goldberg M.D.
Director, Division of Diagnostic Ultrasound,
Jefferson Ultrasound Research and Education Inst.
Thomas Jefferson University,
135 South 10^th Street
Philadelphia, PA 19107
Phone: 215 955 8534
Fax: 215 955 8549
Email: Barry.B.Goldberg@mail.tju.edu

Chris Merritt M.D.
Division of Diagnostic Ultrasound
Jefferson Ultrasound Research and Education Inst.
Thomas Jefferson University
135 South 10^th Street
Philadelphia, PA 19107-555244
Phone: 215 955 8534
Fax: 215 955 8549
Email Crbm@concentric.net

Michael F. Insana, Ph.D.
Professor of Radiology
University of Kansas Medical Center
3901 Rainbow Blvd.
Kansas City, KS 66160-7234
Phone: 913-588-6893
Fax: 913-588-7876
Email: insana@research.kumc.edu

Robert Mattrey, MD
Professor of Radiology
UC San Diego Medical Center,
200 West Arbor Drive
San Diego, CA., 92103
Phone: 619 543 6766
Fax: 619 543 3921
Email: rmattrey@ucsd.edu

Jonathan Ophir, Professor & Director (Unable to attend but copy on correspondence for input)
Ultrasonics Laboratory/ Dept. of Radiology
University of Texas Medical School
6431 Fannin Street, Suite 6.168
Houston, Texas 77030, USA
713-500-7686 (voicemail)
713-500-7687 (secretary)
713-500-7694 (fax)
281-987-0444, ext. 10485 (24-hour pager)
Email: jophir@ieee.org

Dr. Patrick L. Von Behren, Ph.D.
Director, Advanced Ultrasound Engineering,
Siemens Medical Systems, Inc.
22010 SE 51st St.
Issaquah, WA 98029-7002
Tel (425) 392-9180; (425)-557-1348
Fax 425)-557 1779
Email: vbehren@sqi.com

John Allison
Acuson Corporation
1200 Charleston Road.
PO Box 7393
Mountain View, CA 94039-7393
Phone: 650 694 5439
Email: J.Allison@acuson.com

Kei Thomenius
General Electric Corporate Research and Development
PO Box 8, Building KW, RC308B
Schenectady, NY. 12301
Phone: 518-387-7233
Email: thomeniu@crd.ge.com

Helen F. Routh Ph.D. ( Unable to attend but copy correspondance for input)
Chief of Staff
Senior Technology Staff
ATL Ultrasound
Email. Hrouth@atl.com

Alternate: Dr. Patrick Pesque
Director of High End Product Development.
ppesqu@corp.atl.com
Phone: (425) 487 8109.
22100 Bothell Everett Highway
Post Office Box 3003
Bothell, WA 98041-3003

NSF Representatives:

Gilbert B. Devey
Program Director
Division of Bioengineering and Environmental Systems,
National Science Foundation,
National Science Foundation
4201 Wilson Blvd.
Arlington, VA 22230
Email: gdevey@nsf.gov

Dr. John H. Cozzens
Program Director, Signal Processing Systems, Room 1155
National Science Foundation
4201 Wilson Blvd.
Arlington, VA 22230
TEL: 703-306-1914
Email: jcozzens@nsf.gov

DARPA Representatives

Wallace Smith Ph.D.
Project Manager
Defense Science Office
DARPA
3701 N. Fairfax Drive
Arlington, Virginia 2203-1714
Phone: 703-606 0284
Email: wsmith@darpa.mil

Office Of Woman’s Health

Faina Shtern, M.D.
Associate Director for Research and Technology Affairs
PHS Office of Woman’s Health
200 Independence Avenue, S.W.
Washington, DC 20201
Phone: 202-690-7650
Email: FShtern@osophs.dhhs.gov