Background-Free Detection of Light-Activatable Perfluorocarbon Nanodroplets

Perfluorocarbon nanodroplets (PFCnDs) have been studied as an activatable ultrasound contrast agent. During vaporization, light-activatable PFCnDs generate characteristic ultrasound and photoacoustic (US/PA) signals across a sequence of images. As more PFCnDs vaporize in a region of interest, the US contrast increases exponentially while PA contrast decreases similarly. Previous studies have explored background-free imaging algorithms to show the location of the activated PFCnDs, by detecting the slope of exponential signals. In this study, we generate background-free images by identifying exponential trends and signal amplitudes, thus giving the location of PFCnDs and scaling the images by approximate PFCnD concentration. Synthetic US/PA images were generated by simulating a time-series of US/PA images during PFCnD activation events. Tissue background signals (offset) and random noise were also included. The algorithm analyzes intensity values to detect exponential trends using the coefficient of determination
$(\mathbf{R}^{\mathbf{2}})$
, followed by image reconstruction. A pixel-wise pre-filtering was performed to account for offset and noise. Then, the
$\mathbf{R}^{\mathbf{2}}$
coefficient was calculated to evaluate how each pixel's data fit an exponential behavior. Image reconstruction used the parameters of the fitting function generated, using the estimated amplitude value to scale the concentration image. A Pearson correlation coefficient was used to assess the accuracy of the algorithm by comparing the reconstructed image with the ground truth image. The algorithm was tested under presence of up to 70% noise and 50% signal offsets to assess the potential applicability in realistic tissue environments. In these conditions, the algorithm showed strong performance with high correlation coefficients. The successful reconstruction of background-free images suggests the feasibility of this method for further testing in vivo.