Compressive hyperspectral Raman imaging via randomly interleaved scattering projection

Abstract

Recently, compressive sensing has been introduced to confocal Raman imaging to accelerate data acquisition. In particular, unsupervised compressive imaging methods do not require a priori knowledge of an object’s spectral signatures, and they are thus applicable to unknown or dynamically changing systems. However, the current methods based on either spatial or spectral undersampling struggle between spatial and spectral fidelities at high compression ratios. By exciting a sample with an array of focused laser beams and randomly interleaving the projection locations of the scattering, we simultaneously demonstrate a single-acquisition confocal Raman hyperspectral imaging with a high fidelity and resolution in spatial and spectral domains, at a compression ratio of 40–50. The proposed method is also demonstrated with suppressed noise and a smooth transition at the boundaries.

Compressive Raman imaging by combining scattering-projection interleaving with context-aware excitation

“…Same as the original SIRI, the system uses two pairs of galvo mirrors for focus-array generation (GM1 and GM2) and interleaving of Raman scattering (GMx and GMy), respectively. 23 The only difference is that, rather than exciting the entire FOV with a regular two-dimensional (2D) focus-array, context-aware SIRI excites only a smaller ROI covering the interested target with a non-regular focus array. To do this, a bright-field image of the FOV is acquired, and an adaptive threshold segmentation method is used to identify the pixels within the target area (see below for details).…”

“…The optical setup was almost the same as previously reported. 23 A laser at 785 nm was used for Raman excitation (LD785-SEV300, Thorlabs), and a 100× oil immersion objective (RMS100X-PFO, Olympus) was used for bright-field imaging and collecting Raman images. Two pairs of galvo mirrors (GVS002, Thorlabs) were used to steer the laser beam and to project the Raman scattering, respectively.…”

“…In order to reconstruct the original hyperspectral image u from the compressed measurements y , the inverse problem of eqn (1) is solved by converting it into an unconstrained optimization problem:where φ (·) is a regularization function to ensure the smoothness of the data cube in the spatial and spectral domains, and γ is the regularization parameter. In this work, a Hessian penalty function was used to explore the sparsest solution in the second-order gradient space, 23 and eqn (2) was solved using the two-step iterative shrinkage/thresholding (TwIST) algorithms. 29…”

Fast Raman imaging through the combination of context-aware matrix completion and low spectral resolution

“…spatial or spectral sparsity, or that the image matrix is low-rank). [34][35][36][37][38][39][40][41] Several compressive Raman spectral imaging systems, including both spatial and spectral compression 35,37,42,43 have been proposed in recent years. We recently have shown that combining spatial sparsity with morphological priors in a standard Raman imaging system, imaging time can be reduced by 3-to-5-fold.…”

Optical time‐stretch imaging flow cytometry in the compressed domain

“…In recent years, compressed sensing (CS) has been successfully used in many imaging systems to significantly reduce the amount of data sampling, speed up signal acquisition, and alleviate the pressure of data processing. For instance, CS has enabled single-pixel imaging (SPI) [21][22][23], compressed ultrafast photography (CUP) [24][25][26], and compressive hyperspectral Raman imaging [27][28][29]. The success of these CS-based imaging methods relies on the fact that the signal usually contains only a few nonzero values in a particular domain (e.g., Fourier domain [30], discrete cosine domain [31], discrete wavelet domain [32]), and a small number of measurements are sufficient to capture them.…”