Multispectral photoacoustic imaging (PAI) is well-known to be prone to various noise and artifacts. A type of noise in multispectral PAI may be similar to that present in hyperspectral imaging (HSI) such as pattern noise due to calibration error [1], while another type of noise, such as additive electronic noise, comes from system thermal noise or electromagnetic interference [2]. Multispectral PAI artifacts can be due to motion or image reconstruction. Motion-based artifacts arise due to movement of an object such as breathing or heart beat [3]. Reconstruction-based artifacts can arise due to limited angle issues in backprojection reconstruction algorithms. For instance, spatial undersampling can lead to streak artifacts during image reconstruction due to limited number of elements in the transducer array. In these contexts, denoising multispectral PAI and removing artifacts is a crucial procedure for further image processing and analysis such as image segmentation, spectral unmixing or co-registration.
