DEMONSTRATION OF VIDEO-RATE DIFFUSE OPTICAL TOMOGRAPHY IN PHANTOMS AND TISSUES Brian W. Pogue 1* , Daqing Piao 2 , Hamid Dehghani 1,3 , Keith D. Paulsen 1 1 Thayer School of Engineering, Dartmouth College, Hanover NH 03755. 2 Department of Electrical Engineering, Oklahoma State University, Stillwater OK. 3 School of Physics, University of Exeter, Exeter UK. * Email: pogue@dartmouth.edu ABSTRACT Near-infrared diffuse optical tomography has been demonstrated with video rate acquisition of the transmitted signal for 8 sources and 8 detectors. The system design is outlined with components illustrated, and tomographic images are shown for phantoms and tissues. The system uses spectral encoding of the laser sources at small wavelength increments to allow each source location to be discerned at the detector. The key to this is spectrally dispersing the different wavelengths prior to detection, with a video rate CCD. Possible uses for this type of system are in the area of physiological monitoring and contrast agent kinetic imaging. (a) (b) 1. INTRODUCTION Near-infrared imaging with diffuse light in tomography mode has been examined for a couple of decades. Only recently has there been a focus on developing fast imaging systems, as much of the attention has focused on biological applications which did not require fast imaging in tomographic mode. Fast imaging in remission mode has been a stable of functional activation studies, yet in tomographic mode, the ability to resolve location has been problematic, because the resolution of diffuse tomography is quite limited. However in recent years, the marriage of diffuse tomography with ultrasound [1], tomosynthesis [2] and MRI [3-5] have shown that diffuse tomography may not be limited by the low resolution limitation of the inverse image reconstruction problem. Rather, it is possible to couple such systems into the standard clinical imaging modalities and utilize their superior localization capabilities, and then provide NIR spectral information about those localized regions. Figure 1. Graphical representation of two methods to parallelize the source locations (smaller red input arrows) to be measured at a single detector location (larger black output arrow). In (a) different heterodyne frequencies are used to encode the multiple source locations, which can then be detected simultaneously with one detector and the signal mixed with the original carrier frequency, f. Then each source signal is extracted from the Fourier transform of the data, where the location of the source is encoded by δf 1 , δf 2, δf 3 . This approach has the inherent limitation that each additional source reduces the dynamic range of the detector, and hence decreases the signal to noise ratio. In (b), the spectral encoding of the sources is accomplished by offsetting the wavelength by small amounts, δλ 1 , δλ 2, δλ 3 , which are sufficient to allow signal separation at the output by using a high resolution spectrometer. The detection of the signal can still be done in frequency domain if coupled to multiple fast PMTs. In this abstract, we discuss one aspect of an advanced tomography system, namely the ability to design a system which achieves video frame rate [6, 7], similar to an ultrasound system, whereby tomographic data and images are provided by having parallel source and parallel detection simultaneously. There are several configurations which would possibly allow this type of detection, including