Enhancement of the resolution for low-quality fluorescence microscopy image by novel computational method

The application of computational methods for enhancement the resolution of fluorescence microscopy images beyond the diffraction limit has emerged in recent years. Among them, super-resolution radial fluctuations (SRRF) and mean-shift super-resolution (MSSR) are two widely used representatives. However, these two methods are often unsatisfactory when dealing with low-quality fluorescence microscopy images, which are prone to artifacts as well as structure discontinuities. Here, we propose an effective computational method named morphological filtering and polynomial fitting (MFPF). MFPF equips excellent denoising and light-field balancing capabilities, which are beneficial for subsequent resolution enhancement. Our results validated on public BioSR dataset show that MFPF not only improves the quality of subcellular structure images such as microtubules and endoplasmic reticulum, but also attenuates the artifacts and discontinuities generated by SRRF radiality map analysis and single-frame MSSR reconstructions while maintaining the comparable resolution. Furthermore, it permits processing of low-quality fluorescence microscopy images, and would have vast applications in long-term super-resolution imaging of live cells with moderate image quality.