We propose a novel noniterative near-infrared diffusive image reconstruction method that uses minimal a priori co-registered ultrasound information. Small absorbing targets embedded in a homogeneous background are described approximately in terms of their monopole, dipole, and quadrupole moments. With an approximate estimation of the center locations of these absorbers from ultrasound images, we show in simulations that the reconstruction accuracy of the absorption coefficient exceeds 80% if the noise level is less than 0.2%. We also demonstrate experimentally that the accuracy can be improved by use of additional ultrasound volume information even for a noise level as high as 1.5%.
展开▼
