superresolution photoacoustic imaging

An imaging comparison of blood flowing through five channels using various approaches. At the top are single photoacoustic images from the analyzed image stack. The image at the bottom left is the result of conventional analysis; the middle- and right-bottom images were obtained when the researchers used their new fluctuation-based analysis approach, with five channels clearly resolved in the final fluctuation analysis. [Image: Bastien Arnal / Grenoble University, France] [Enlarge image]

Photoacoustic imaging uses light and ultrasound to view structures beyond the depths of what optical microscopy alone can capture. However, photoacoustic imaging is limited by acoustic diffraction, which occurs at wavelengths too long to resolve microscopic blood vessels and capillaries below the skin’s surface. An international team of researchers led by Ori Katz, Hebrew University of Jerusalem, Israel, reports testing a new approach to super-resolution photoacoustic imaging that uses math—not extra equipment—to break the acoustic diffraction limit (Optica, doi: 10.1364/OPTICA.4.001397). The researchers say that the technique could someday be used for real-time label-free in vivo cellular imaging.

Old equation, new problem

The key to the Katz team’s new approach is a nearly decade-old statistical method that was originally used to improve the resolution of fluorescence microscopy. The method, called super-resolution optical fluctuation imaging (SOFI), uses changes in light absorption from moving elements in the imaged sample to computationally increase spatial resolution.

Specifically, the researchers used an advanced SOFI-based nth-order statistical temporal fluctuation analysis in two proof-of-concept demonstrations to calculate super-resolved photoacoustic images of microbeads and red blood cells (RBCs) as they flow through vessel-like microchannels. By relying on light-absorption data, Katz and his colleagues were able to exploit light’s superior diffraction limit, which is two orders of magnitude better than ultrasound, to greatly improve photoacoustic-imaging resolution.

Pushing the acoustic limits of diffraction

The team’s proof-of-concept setup included five parallel microchannels positioned perpendicular to the imaging plane of a linear ultrasound array. An ultrasound scanner captured 2-D photoacoustic images of flowing microbeads or RBCs. The microbeads were 10 µm in diameter (slightly larger than the average 6.2-to-8.2-µm-wide RBC). Both the microbeads in solution and whole human blood moved through the microchannels at a realistic in vivo flow rate of 25 mm/s.

As the samples flowed through the microchannels, they were homogenously illuminated with 5-ns laser pulses at a 20-Hz repetition rate from a frequency-doubled Nd:YAG laser. Photoacoustic signals were acquired for each laser pulse, which corresponded to one “frozen” random distribution of the microbeads or RBCs in the channels.

The team reconstructed images from the acoustic signals using a conventional delay-and-sum algorithm. To resolve the five different channels, the researchers conducted nth-order cumulated analyses. The researchers were also able to eliminate imaging artifacts by extending the SOFI framework to process complex-valued acoustic signals.

Validating the resolution increase

The researchers reported resolution values beyond the acoustic diffraction limit from the proof-of-principle demonstrations. These values closely resembled the predicted resolution increases calculated using numerical simulations.

During the whole-blood demonstration, the researchers found that it was “considerably more difficult to insure a stable and controllable behavior of the blood flowing through the channels.” However, the results showed that by analyzing a specific data set with more than 15,000 images taken over 2.5 minutes, it was possible for them to calculate images beyond the acoustic diffraction limit.

The authors say the whole-blood proof-of-concept demonstration suggests that their method could allow scientists to view, in real time, blood vessels and blood flow in vivo with existing imaging hardware. They say that the main hurdle to introducing this approach to in vivo applications is the randomness of RBC distribution and the motion-induced fluctuations in living samples and more complicated imaged structures.