CNN-based TEM image denoising from first principles

2025-01-20

AI Summary

Deep learning-based denoising of transmission electron microscopy (TEM) images is hindered by complex, physically heterogeneous noise and the absence of ground-truth data. Method: We propose a first-principles-driven paradigm: (i) high-fidelity ground-truth TEM images are synthesized via density functional theory (DFT) coupled with pseudo-atomic orbital basis sets; (ii) a physically accurate training dataset is constructed by systematically modeling four realistic noise sources—shot noise, readout noise, quantization noise, and detector modulation transfer function effects; (iii) a lightweight, noise-type-specific CNN architecture is designed for robust cross-intensity denoising. Contribution/Results: Our method achieves significant improvements in peak signal-to-noise ratio (PSNR) and atomic structure fidelity, maintaining stable performance on unseen noise levels. Furthermore, it uncovers critical limitations—including circular distortion and stitching artifacts—providing concrete physical insights to guide future physics-informed model refinement.

Technology Category

• Machine Learning: Deep Neural Architectures and Foundation Models
• Computer Vision: Other Foundations of Computer Vision
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding

Application Category

• Security and Privacy: Data transparency and provenance
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Methods, algorithms and applications for the development of semantic models, knowledge graphs and other forms of structured data models with machine-interpretable semantics

Abstract

Transmission electron microscope (TEM) images are often corrupted by noise, hindering their interpretation. To address this issue, we propose a deep learning-based approach using simulated images. Using density functional theory calculations with a set of pseudo-atomic orbital basis sets, we generate highly accurate ground truth images. We introduce four types of noise into these simulations to create realistic training datasets. Each type of noise is then used to train a separate convolutional neural network (CNN) model. Our results show that these CNNs are effective in reducing noise, even when applied to images with different noise levels than those used during training. However, we observe limitations in some cases, particularly in preserving the integrity of circular shapes and avoiding visible artifacts between image patches. To overcome these challenges, we propose alternative training strategies and future research directions. This study provides a valuable framework for training deep learning models for TEM image denoising.

Problem

• Transmission Electron Microscopy
• Image Noise
• Image Analysis

Innovation

• Deep Learning
• CNN
• TEM Image Denoising