Charles A. DiMarzio- dimarzio@ece.neu.edu
440 Dana Building, Northeastern University, 360 Huntington Avenue
Boston, MA 02115
Popular version of paper 4pBB2
Presented Thursday afternoon, Nov 18, 2004
148th ASA Meeting, San Diego, CA
This paper discusses a technique for imaging inside the human body using a combination of light and ultrasound. Many different technologies have been brought to bear on the challenge of imaging the human body. Each has its depth of penetration, spatial and temporal resolution, and contrast mechanism related to medical conditions of interest. Although light is probably our most common imaging modality in everyday life, it is so strongly scattered by tissue that it has not found much application in imaging the body, other than the eye. The use of light is attractive because the absorption of light by blood varies with wavelength of the light. Furthermore, the absorption depends on whether oxygen is bound to the hemoglobin molecules in the blood, so the potential exists to image the amount of blood and the oxygen saturation, which are related to metabolism, and thus potentially useful for medical diagnosis.
In the past decade, advances in lasers and computing have opened the door for an optical imaging technique called diffusive optical tomography. Light is passed through the area of interest from multiple sources to multiple detectors, and the small differences among different combinations are used in a computer to construct an image. The major problem is that resolution is limited to many millimeters at useful depths. One can observe the problem by placing one's thumb over a red light-emitting diode, such as is found on many electronic displays. While light is clearly transmitted through the thumb, scattering prevents formation of a clear image of the internal structure of the thumb. The problem is even worse when imaging a breast, or a fetus in utero. Currently researchers in the field of inverse problems are trying to improve the resolution with better computational techniques.
To complement that work, our project modifies the imaging technique by the introduction of ultrasound. As the light passes through the body, it scatters randomly from different tissue components. Light that travels over different paths will arrive at different times. When light waves arrive with their peaks and valleys synchronized in time, they add to produce a bright region. When the peaks of one are synchronized with the valleys of another, they tend to cancel, producing a darker region. The resulting random distribution of light and dark is called a speckle pattern. Introducing ultrasound into the paths of the light modifies the path lengths, altering the speckle pattern at the frequency of the ultrasound. Thus, at a particular point, the brightness increases and decreases at this frequency. We know then that the light which varies in this way must have passed through the ultrasound wave. Because the ultrasound wave can be focused to a small spot, we can use this information to improve the spatial resolution of the optical measurement. Furthermore, we can move the ultrasond wave to different locations to cover the region of interest.
This technique has three major problems. (1) The signals are small. (2) Averaging over a large enough region in an attempt to increase the signal just smooths out the variations of interest. (3) The ultrasound wave is a focused beam, much like a laser beam, and the modulated light may have travelled through any part of the beam, and not just the focus.
This talk will focus on a new approach, developed by a team at CenSSIS, the Center for Subsurface Sensing and Imaging Systems, at Northeastern University and Boston University. While the author, at Northeastern, was working on the interaction of light and ultrasound, Prof. Todd Murray of BU was working with photorefractive crystals to detect surface motion for another medical imaging technique. His approach is to use this crystal to distort a wave of laser light so that it matches the speckle pattern, and mix it with the speckle pattern modulated by the surface motion. We decided to combine the two techniques, using the photorefractive crystal to detect the ultrasound modulation of light.
In experiments conducted in Prof. Murray's laboratory, we found not only the expected increase in signal strength, but three unexpected results. First, the signal is "demodulated," so that it is more easily detected. Second, the demodulation is "coherent" so that light can be detected over a larger area without smoothing the desired signal. Finally, the effect is strongest at the focus of the ultrasound, so the path of the light is more tightly localized.
We continue to conduct experiments using fabricated "tissue phantoms" which have optical and ultrasound properties similar to those of tissue, and we continue mathematical analysis to better understand the effect. We anticipate that, with further research, this technique can be developed sufficiently to be used in a variety of applications in medical imaging. The author wishes to thank the team of faculty and graduate students who performed this research, including Todd W. Murray, Ronald A. Roy, Lei Sui, and Gopi Maguluri, all at BU; Alex Nieva and Florian Blonigen, at Northeastern. This work was supported in part by CenSSIS, the Center for Subsurface Sensing and Imaging Systems, under the Engineering Research Centers Program of the National Science Foundation (award number EEC-9986821).