An optimized locally adaptive non-local means denoising filter for cryo-electron microscopy data

作者：Wei Dai Yu; Yin Chang Cheng^*

来源：Journal of Structural Biology, 2010, 172(3): 211-218.

DOI：10.1016/j.jsb.2010.06.021

• Cryo-electron microscopy
• Image processing
• Noise reduction
• Non-local means filter

Cryo-electron microscopy (cryo-EM) now plays an important role in structural analysis of macromolecular complexes, organelles and cells. However, the cryo-EM images obtained close to focus and under low dose conditions have a very high level of noise and a very low contrast, which hinders high-resolution structural analysis. Here, an optimized locally adaptive non-local (LANL) means filter, which can preserve signal details and simultaneously significantly suppress noise for cryo-EM data, is presented. This filter takes advantage of a wide range of pixels to estimate the denoised pixel values instead of the traditional filter that only uses pixels in the local neighborhood. The filter performed well on simulated data and showed promising results on raw cryo-EM images and tomograms. The predominant advantage of this optimized LANL-means filter is the structural signal and the background are clearly distinguishable. This locally adaptive non-local means filter may become a useful tool in the analysis of cryo-EM data, such as automatic particle picking, extracting structural features and segmentation of tomograms.

• 出版日期2010-12
• 单位北京大学