Structure-dependent amplification for denoising and background correction in Fourier ptychographic microscopy

Abstract

Fourier Ptychographic Microscopy (FPM) allows high resolution imaging using iterative phase retrieval to recover an estimate of the complex object from a series of images captured under oblique illumination. FPM is particularly sensitive to noise and uncorrected background signals as it relies on combining information from brightfield and noisy darkfield (DF) images. In this article we consider the impact of different noise sources in FPM and show that inadequate removal of the DF background signal and associated noise are the predominant cause of artefacts in reconstructed images. We propose a simple solution to FPM background correction and denoising that outperforms existing methods in terms of image quality, speed and simplicity, whilst maintaining high spatial resolution and sharpness of the reconstructed image. Our method takes advantage of the data redundancy in real space within the acquired dataset to boost the signal-to-background ratio in the captured DF images, before optimally suppressing background signal. By incorporating differentially denoised images within the classic FPM iterative phase retrieval algorithm, we show that it is possible to achieve efficient removal of background artefacts without suppression of high frequency information. The method is tested using simulated data and experimental images of thin blood films, bone marrow and liver tissue sections. Our approach is non-parametric, requires no prior knowledge of the noise distribution and can be directly applied to other hardware platforms and reconstruction algorithms making it widely applicable in FPM.

Optical mesoscopy, machine learning, and computational microscopy enable high information content diagnostic imaging of blood films

“…With a camera exposure time of 100 ms, the total acquisition time for each (monochrome) image was slightly less than 30 s. Colour images were captured by combining images acquired under illumination by red, green, and blue LEDs. Images were reconstructed using a version of the iterative phase retrieval method described by Tian et al [13] modified to reduce background‐related image artefacts [14].…”

Histological and cytological imaging using Fourier ptychographic microscopy

“…For FPM imaging of peripheral blood films we used a 10x/0.3 objective lens, a sample-LED distance of 62 mm (corresponding to a passband overlap of 75% and 𝑁𝐴 1)2 of 0.9). DF background signals were corrected using the method described by Claveau et al 7 . For imaging histological tissue sections we used a 4x/0.16 objective, a sample-LED distance of 87 mm (corresponding to a passband overlap of 69% and 𝑁𝐴 1)2 of 0.7).…”

“…If the subtracted level is too low the orange peel artefact is not fully removed, if it is too high real sample information is removed from the DF images which reduces the spatial resolution and contrast of the reconstructed image. For imaging relatively sparse, well separated objects (such as cells in a thin peripheral blood film) we have found that a nonparametric 'structure dependent amplification' method is effective 7 . In this approach data redundancy in real space is exploited to create maps to retain useful information in DF images whilst suppressing the background.…”

Improved reconstruction with Butterworth-weighted transfer function for Fourier ptychographic microscopy

“…For the second kind, in order to overcome the limitation of the traditional GS algorithm, more algorithms [11,12] have been proposed. Moreover, some anti-noise or denoising algorithms [13][14][15] are applied taking the existed noises into consideration. With regard to determine the position of each sub-frequency domain, the simulated annealing algorithm, the quasi Newton algorithm, and the nonlinear regression analysis algorithm are employed to calibrate the position error [16][17][18][19].…”