Brightfield image deconvolution

Brightfield image deconvolution was conducted using the Lucy-Richardson⁸⁵ algorithm in MATLAB. Before starting the deconvolution, the point spread function (PSF) of the smartphone microscope was measured by imaging single 500-nm polystyrene microbeads in brightfield mode. To obtain the PSF image, 20 single microbeads were averaged. The averaged image of a single microbead was then split into red (R), green (G) and blue (B) channels (Supplementary Fig. 1). The central 2 by 2 pixels were used as the PSF of each color channel during the deconvolution process. To deconvolve raw images, the original smartphone brightfield images were also split into R, G and B channels. Then, each channel was deconvoluted separately with the corresponding PSF of that channel. The deconvoluted single channel images were then merged in ImageJ to generate the deconvoluted RGB image with enhanced resolution and contrast. The intensity value of each deconvoluted color channel was manually adjusted in ImageJ to compensate the color difference before and after deconvolution.