Comparison of noise reduction methods in photoacoustic microscopy.

Photoacoustic microscopy (PAM) is classified as a hybrid imaging technique based on the photoacoustic effect and has been frequently studied in recent years. Photoacoustic (PA) signals are inherently recorded in a noisy environment and are also exposed to noise by system components. Therefore, it is essential to reduce the noise in PA signals to reconstruct images with less error. In this study, an image reconstruction algorithm for PAM system was implemented and different filtering approaches for denoising were compared. Studies were carried out in three steps: simulation, experimental phantom and blood cell studies. FIR low-pass and band-pass filters and Discrete Wavelet Transform (DWT) based filters (mother wavelets: "bior3.5″, "bior3.7″, "sym7″) with four different thresholding techniques were examined. For the evaluation purposes, Root Mean Square Error (RMSE), Signal to Noise Ratio (SNR) and Contrast to Noise Ratio (CNR) metrics were calculated. In the simulation studies, the most effective methods were obtained as: sym7/heursure/hard thresh. combination (low and medium level noise) and bior3.7/sqtwolog/soft thresh. combination (high-level noise). In experimental phantom studies, noise was classified into five levels. Different filtering approaches perform better depending on the SNR of PA images. For the blood cell study, based on the standard deviation in the background, sym7/sqtwolog/soft thresh. combination provided the best improvement and this result supported the experimental phantom results.