Sound Speed Tomography for Early Breast Cancer Detection

Institution: University of California, San Diego
Investigator(s): Jakob Nebeker, B.S. -
Award Cycle: 2009 (Cycle 15) Grant #: 15GB-0023 Award: $74,325
Award Type: Dissertation Award
Research Priorities
Detection, Prognosis and Treatment>Imaging, Biomarkers, and Molecular Pathology: improving detection and diagnosis

Initial Award Abstract (2009)

Early detection is the most important factor determining survival from breast cancer. The goal of this project is to improve early detection of cancer; especially in dense breasts that mammography has difficulty in assessing. We hope to help earlier breast cancer detection by producing tomographic (i.e., serial, thin sections through a tissue) sound speed images using our Volume Breast Ultrasound (VBUS) scanner. Contrary to standard ultrasound, which measures sound reflection for each point in the image, sound speed images capture the tissue sound speed at each point. It has been previously shown that sound speed varies greatly in tissues, and in particular breast tumor tissue has a much higher sound speed than healthy tissue. There is also a strong relationship between sound speed and the temperature of tissue.

Our previously constructed VBUS scanner consists of a table, a water bath, a robotic gantry system, a transducer, an ultrasound system, and a PC workstation for acquisition and reconstruction. We will use this system for the sound speed imaging. To generate sound speed images we will use the back-projection of measured time of flight (ToF) for each ultrasound A-line “ray”. Time of flight will be measured by examining the back wall of the scanning tank. Back-projecting time of flights over 360 degrees produces an image of “slowness”, which can be inverted to create a sound speed image. We will design and test a novel, iterative, “bent-ray” back-projection algorithm to account for sound refraction in the tissue. The ultrasound system is a Terason T3000 that allows us direct access to the radio-frequency (RF) echo data for each A-line as the B-mode image is being constructed. We will use this higher-resolution RF data to perform the sound speed reconstruction. We will test our sound speed detection system using artificial breast phantoms, especially under conditions of temperature variability that would mimic the situation in breasts with embedded tumors.

High-resolution sound speed images have potential to help in early breast cancer detection for women with dense breasts. Furthermore, real-time sound speed images would allow monitoring and control of high high-intensity focused ultrasound (HIFU) treatment of breast cancer.

Final Report (2011)

While X-ray mammography is the current "gold standard" for early detection of breast cancer, it has drawbacks including exposing women to ionizing radiation, causing discomfort, and having reduced sensitivity in dense breasts. Ultrasound is non-ionizing, uses no compression, and has high contrast even in dense breasts but has been held back due to non-repeatability and high operator dependence. To ameliorate the disadvantages we have built a volume breast ultrasound scanner (VBUS) that is repeatable, reliable, and produces images of higher resolution and contrast than conventional B-Mode ultrasound.

The purpose of this project was to enhance the VBUS scanner to produce high-resolution “sound speed” images and to evaluate those images under various applications. Because the sound speed of tissue varies considerably, with cancer tissues showing higher sound speed compared to healthy breast tissue, these sound speed images could prove to be of great help in detecting and diagnosing breast cancer.

Progress towards project specific aims:

  1. Develop and refine the sound speed imaging algorithm. We have continued to develop and refine our sound speed imaging method using high quality RF data as well as measured the effects of diffraction of a pulse echo beam using a novel pulse echo sound beam imaging technique. We scanned 13 healthy volunteers under IRB review.
  2. Evaluate the ability to measure accurate sound speed in breast tissue phantoms. We have measured quantitative results of our sound speed images using a series of custom test objects and compared results to a time-of-flight velocimeter. The mean error comparing sound speed image data to the time-of-flight velocimeter measurements was 0.6±0.8% with a range of -0.3% to 1.7%. A series of water sound speed measurements made over a temperature range from 26 to 68 °C showed 1% agreement between scanner performance and theory. Performance in acrylamide test objects with diameters ranging from 1.5 to 15 mm showed detectability of a 1% sound speed difference in a 2.4 mm cylindrical inclusion with a CNR of 7.9 dB.
  3. Evaluate the ability to monitor temperature change in a breast phantom. Sound speed tomogram images were calibrated to thermocouple data making it possible to measure temperature in phantoms and potentially volunteers. Mapping of temperature distributions as an acrylamide phantom was heated and cooled based on images acquired every 30 seconds showed a linear relationship (temperature = 0.6392(speed) -938.6) between sound speed and temperature (r=0.95) with standard deviation of error of 0.52 °C over a range of 25-60 °C.

Symposium Abstract (2010)

Jacob Nebeker Jacob and Thomas Nelson

The goals of this dissertation (JN) and IDEA (TN) projects are to improve early detection of breast cancer by building an ultrasound scanner that can image the entire breast. We also are incorporating capability to measure sound speed and attenuation in addition to reflectivity we can gain additional diagnostic information regarding breast lesions further improving diagnostic accuracy. Because breast tumors have been shown to have a markedly higher sound speed than healthy tissue, sound speed has the potential to be a valuable new indicator of cancer. Standardizing ultrasound breast imaging provides high quality images improving detection of non-palpable breast cancers that cannot be seen with mammography in women at high risk of breast cancer, especially in women with dense breasts.

Specifically, our goal was to design, construct and begin testing a dedicated volume breast ultrasound (VBUS) scanner for the entire breast. To date we have successfully designed the scanner, a scanning table for the pendant breast and the necessary imaging and computer systems to obtain volume breast images. The scanner uses no compression, is fully automated and makes a complete scan in less than 30 seconds under operator control. We acquire approximately 150 B-mode images completely around the breast. Spatially compounding the images creates a high-detail reflection image. We also obtain reflected data to create sound speed and sound attenuation images. Back-projecting the time delays and attenuation along each ray over all angles produces sound speed and attenuation images.

We have performed a series of measurements to characterize scanner performance with satisfactory results. Gel wax test object data that we can distinguish lesions down to approx 1mm in diameter in the reflection images and approx 3mm in size in the sound speed and attenuation images. Reconstructed sound speed contrast resolution has been measured to be approx 0.3% (1500 m/s +-5 m/s) for synthetic phantoms and 1% (1460 m/s +- 15 m/s) for gel wax phantoms). We also have obtained images from normal volunteers as part of validating performance and determining that the scanner is ready for clinical evaluation.  We are preparing to expand our clinical imaging to a broader clinical trial.

The volume breast ultrasound (VBUS) scanner utilizes a novel imaging technology to improve breast cancer detection and diagnosis. First, the scanner obtains a volume data set for the entire breast improving visualization of breast tissue and providing more precise localization of suspicious breast lesions. Second, the scanner provides a standardized scanning environment reducing much of the variability present in current ultrasound scans leading to more accurate diagnosis and a better prognosis. Third, the scanner uses volume imaging and spatial compounding to reduce speckle and improve lesion conspicuity. Fourth, we simultaneously image sound reflection, speed, and attenuation, and finally the scanner does not use uncomfortable compression and ionizing radiation. We believe this approach has the potential to improve on the state of the art for early breast cancer detection, especially in women with dense breasts.

Imaging of Sound Speed Using Reflection Ultrasound Tomography
Periodical:Journal of Ultrasound Medicine
Index Medicus:
Authors: Nebeker J, Nelson TR.
Yr: 2012 Vol: 31 Nbr: 9 Abs: Pg:1389-1404