Influence of skin tone on target size detectability in photoacoustic breast imaging

Photoacoustic imaging has the potential to improve non-invasive breast cancer diagnosis through the detection of vascular structures. However, illumination through skin introduces a skin tone bias, as greater melanin content increases optical absorption and can create acoustic clutter, reducing the visibility of various target sizes. The impact of skin tone bias as a function of target size was investigated when implementing a fast Fourier transform (FFT)-based reconstruction and short-lag spatial coherence (SLSC) beamforming. These two methods were applied to channel data from multidomain simulations with two optical wavelengths (800 nm and 1064 nm), three tan skin tones with individual typology angles (ITAs) ranging 12° to 26° to assess the impact on 11 spherical targets (with diameters ranging 0.5 mm to 3 mm), as well as one dark skin tone (ITA = -45°) and one very light skin tone (ITA = 60°) to assess outcomes with two vessel structure targets. Each of the investigated targets were embedded in a previously validated realistic 3D breast model. The signal-to-noise ratio (SNR) and generalized contrast-to-noise ratio (gCNR) were measured to quantitatively assess target visibility. With 800 and 1064 nm wavelengths, targets underlying tan skin tones were difficult to visualize with FFT beamforming, with mean SNR and gCNR ≤4.51 and ≤0.61, respectively. SLSC beamforming enabled visualization of simulated target sizes ranging 1.25-3 mm, with mean SNR ≥3.47 and mean gCNR ≥0.57. These improvements translated to improved visualization of simulated vessel structures derived from in vivo photoacoustic images with SLSC beamforming (relative to the FFT reconstruction method). Results are promising to equitably advance next-generation photoacoustic imaging systems for breast cancer diagnosis, considering multiple skin tones.