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

Area of Science:

• Microscopy
• Optical Imaging
• Image Reconstruction

Background:

• Fourier Ptychographic Microscopy (FPM) enables high-resolution imaging via iterative phase retrieval.
• FPM is sensitive to noise and background signals from brightfield and darkfield (DF) images.
• Inadequate DF background removal causes significant artifacts in FPM reconstructions.

Purpose of the Study:

• To investigate noise sources impacting FPM image quality.
• To propose an improved method for FPM background correction and denoising.
• To demonstrate the method's effectiveness in enhancing reconstructed image quality and speed.

Main Methods:

• Utilizing data redundancy in real space to improve DF signal-to-background ratio.
• Optimally suppressing background signals before reconstruction.
• Incorporating differentially denoised images into the FPM iterative phase retrieval algorithm.

Main Results:

• The proposed method significantly outperforms existing techniques in image quality, speed, and simplicity.
• Efficient removal of background artifacts is achieved without suppressing high-frequency information.
• High spatial resolution and sharpness are maintained in reconstructed images.

Conclusions:

• Inadequate DF background correction is a primary cause of FPM artifacts.
• The novel FPM background correction and denoising method offers superior performance.
• The non-parametric approach is widely applicable across FPM platforms and algorithms.