Superresolution photoacoustics

In 1880 Alexander Graham Bell discovered that shining light on a material elicits sound from it. His discovery, known as the photoacoustic effect, has today matured into a sensitive medical imaging technique: The absorption of visible or near-IR light pulses by molecules under the skin increases their temperature and launches a pressure wave that propagates out of the body as ultrasound, which is then picked up by an array of microphones. (See the article by Stanislav Emelianov, Pai-Chi Li, and Matthew O’Donnell, Physics Today, May 2009, page 34 .) The scattering mean free path of ultrasound in soft tissues is on the scale of meters, which makes it straightforward to reconstruct an image of the absorber by triangulating from the microphone array to the acoustic source. The rub is that the resolution of a photoacoustic image is limited by the acoustic wavelength.

Researchers led by Ori Katz (Hebrew University of Jerusalem) and by Emmanuel Bossy (University of Grenoble Alpes in France) have now demonstrated how to improve that resolution beyond the diffraction limit of ultrasound. The trick was to adapt an existing mathematical technique known as superresolution optical fluctuation imaging to photoacoustics. SOFI reconstructs a map of a region by statistically analyzing, pixel by pixel, tens of thousands of photoacoustic snapshots. So long as the absorptive features fluctuate over time between each snapshot, reconstructions of the standard deviation of the pixels’ intensity sharpen the resolution by a factor of √2. Reconstructions based on higher-order statistics sharpen the images yet further—up to a factor of √n for an nth-order analysis.

Red blood cells, which are fast moving and absorb visible light strongly, represent an ideal test case. The researchers imaged a cross-sectional slice of cells streaming through a series of parallel microfluidic channels, each separated by 180 µm, as shown in the top panel. From roughly 15 000 snapshots of a central plane (red), they calculated the mean of a conventional photoacoustic image (middle), in which the channels blur together, and a sixth-order statistical analysis (bottom), which resolves the blood in each of them. The technique, which can be used with current imaging hardware, exploits naturally existing fluctuations—no outside contrast agents required. (T. Chaigne et al., Optica 4, 1397, 2017 .)