Super-resolution stimulated Rama scattering microscopy with adam optimization-based pointillism deconvolution (A-PoD)

1 . A deconvolution method for generating a series of super-resolved images, the deconvolution method comprising:

applying an Adam optimization-based Pointillism Deconvolution (A-POD) algorithm to an image; and

generating a series of super-resolved images.

2 . The deconvolution method of claim 1 , wherein the deconvolution method is effective to enhance special resolution of an image at a high processing speed.

3 . The deconvolution method of claim 1 , wherein the series of super-resolved images enables examination of protein and lipid distribution at a nanoscopic level.

4 . The deconvolution method of claim 3 , wherein the examination of protein and lipid distribution involves direct visualization of lipid metabolic changes in cells or brain tissue.

5 . The deconvolution method of claim 1 , wherein the A-Pod algorithm is configured for use with one or more gradient descent optimization algorithms.

6 . The deconvolution method of claim 1 , wherein generating the series of super-resolved images includes removing one or more low frequency signals.

7 . The deconvolution method of claim 6 , wherein the one or more low frequency signals are removed using spline or Fourier filtering methods.

8 . The deconvolution method of claim 1 , further comprising automatically estimating one or more virtual emitters based on a measured characteristic of the image.

9 . The deconvolution method of claim 8 , wherein the measured characteristic of the image is a variance and/or a mean intensity of the image.

10 . The deconvolution method of claim 1 , wherein the image is derived from photoacoustic microscopy, infrared microscopy, and/or phase contrast microscopy.

11 . The deconvolution method of claim 10 , wherein the super-resolved images enable determination of a point spread function of the image.

12 . The deconvolution method of claim 1 , wherein a range in which the A-Pod algorithm is applied is based on a frequency domain of the image.

13 . A method of processing cellular images, the method comprising:

interpolating cellular images along an optical axis;

resampling the cellular images; and

optimizing the cellular images with an A-POD algorithm,

wherein resulting cellular images are deconvoluted cellular images.

14 . The method of claim 13 , wherein the A-POD algorithm is customizable according to a pixel size of the cellular images.

15 . The method of claim 13 , further comprising:

counting lipid droplets of the deconvoluted cellular images.