Hybrid Denoising and Super-Resolution Approach for EFTEM Tomographic Image Quality Enhancement

Electron microscopy, specifically Energy-filtered Transmission Electron Microscopy (EFTEM), is used to visualize subcellular structures at the nanometric scale. However, EFTEM images often suffer from noise and low resolution, which can obscure important biological details and complicate alignment and 3D reconstruction. In this paper, we propose a hybrid image processing approach that combines nonparametric denoising based on Contourlet Transform with Sharp Frequency Localization (Contourlet-SD) and Deep Wavelet Coefficients Prediction for Super-Resolution (DWSR) to improve the quality of EFTEM tomographic images. The proposed hybrid denoising and super-resolution method remarkably enhanced the quality and resolution of EFTEM tomographic images. This enhanced image quality is anticipated to make alignment easier and improve subsequent 3D modeling accuracy. Metrics including Learned Perceptual Image Patch Similarity (LPIPs), Blind/Referenceless Image Spatial Quality Evaluator (BRISQUE), and Natural Image Quality Evaluator (NIQE) demonstrate notable enhancements based on quantitative analysis. These enhancements demonstrate the method's ability to reveal fine structural details that were previously hidden by noise and resolution constraints.