Acousto-optic imaging is a multi-modal imaging technique where coherent light diffusing in a complex medium is ‘tagged’ over time by a ballistic ultrasound pulse of frequency ωus. The photons which paths cross with the ultrasound pulse undergo the acousto-optic effect, resulting in the frequency shift of ωus that can be selectively detected using heterodyne interferometry. Since the ultrasounds propagate at a known velocity, a time-to-space map of the tagged photons results in an image I(x, z), where x is the lateral direction and z the depth direction of the diffuse medium. Images at propagation depths much greater than the average mean free path, typically ~1mm in biological tissue, can be obtained. In most images obtained so far, the ultrasounds are focused line after line to recover an image, and therefore limited by the probe emission rate which is ~1-10 KHz depending on the probe size and the acoustic pulse power. Therefore, in order to acquire acoustic images at frame rates greater than 1 Hz for ‘direct visualization’ of the system under study, it is crucial to minimize the number of individual acquisitions necessary to reconstruct an image.
Here, we present an alternative probe configuration where plane waves emitted at various angles are used rather than focused waves to tag the diffuse light. This approach, first proposed by P.Kuchment and L.Kunyansky (2010), is similar to X-ray tomography since the image information is contained in the various angular scans performed for one acquisition. Because the piezo-elements on the acoustic probe are non-isotropic emitters, the angular scan is typically limited to +/20 degrees, which is sufficient to recover information and can be improved using more than one probe. An inversion algorithm based on inverse Radon-transform is than used to reconstruct the image