Robust Denoising of OR-PAM Image by Frequency Decomposition

We spot the insufficiency of a band-pass filter (BPF) to recognize the overlapping noise that limits the gain in signal-to-noise ratio (SNR) and produces a low contrast-to-noise ratio (CNR) image on the optical-resolution photoacoustic microscopy system (OR-PAM). To alleviate the issue, frequency decomposition was utilized to further decompose the frequency spectrum within the band-passed frequency of multiple A-lines PA signals (e.g., during C-mode imaging) into principal spectra (PS) based on the singular value decomposition (SVD) technique. We hypothesized that the PA signal could be recovered from noise by selecting the appropriate PSs. We found PSs carry better decomposing performance that splits signal and noise beyond BPF. With this property, higher CNR and doubled SNR were achieved.