In vivo photoacoustic breast imaging and the interactive effects of skin tone, wavelength, beamformer, and depth on target detectability

Abstract Photoacoustic imaging is promising for breast cancer diagnosis. However, image quality diminishes with greater melanin content with delay-and-sum (DAS) beamforming, which can be resolved with short-lag spatial coherence (SLSC) beamforming. We investigated the applicability of SLSC beamforming to photoacoustic breast data acquired from 29 patients with skin tone individual typology angles (ITAs) ranging −78° (dark) to 71° (very light), acquired with 757 nm and 1064 nm wavelengths. In shallow targets (<5 mm), with 1064 nm wavelength, target detectability (defined as a scaled average of generalized contrast-to-noise and signal-to-noise ratios) decreased from a mean of 0.80 with DAS beamforming to a mean of 0.33 with SLSC beamforming ( p < 0.05). Shallow targets were not visible with 757 nm wavelength when ITA < 14°. In deeper targets, when ITA ≤ 0°, the mean target detectability aggregated across both wavelengths was 0.74 and 0.85 with DAS and SLSC beamforming, respectively. Results were validated with simulated and/or phantom data, and statistical significance was demonstrated with a linear mixed-effects model ( p < 0.05), providing the first known tripartite analysis to assess competing interactions among beamformers, wavelengths, and target depths in clinical photoacoustic imaging.