F-ANcGAN: An Attention-Enhanced Cycle Consistent Generative Adversarial Architecture for Synthetic Image Generation of Nanoparticles

2025-05-23

AI Summary

Nanomaterial analysis suffers from a severe scarcity of high-quality annotated data, hindering accurate segmentation of nanoparticle topological structures. To address this, we propose an attention-enhanced cycle-consistent generative adversarial network (Att-CycleGAN) that synthesizes high-fidelity scanning electron microscopy (SEM) images using only a small number of annotated samples. Our method innovatively integrates a self-attention mechanism with a Style U-Net architecture for the generator and couples it with a U-Net-based segmentation network to enable structural-aware, cross-domain collaborative training. Additionally, a lightweight post-processing strategy is introduced to further enhance image realism. Evaluated on a TiO₂ nanoparticle dataset, Att-CycleGAN reduces the Fréchet Inception Distance (FID) significantly—from 17.65 to 10.39—and consistently improves downstream segmentation performance. This work effectively alleviates the small-sample bottleneck in nanoscale imaging analysis.

Technology Category

• Computer Vision: Generative Adversarial Networks (GANs) for Vision
• Natural Language Processing: Generation
• Machine Learning: Deep Generative Models & Autoencoders

Application Category

• Web Mining and Content Analysis: Web data generation and simulation
• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs

Abstract

Nanomaterial research is becoming a vital area for energy, medicine, and materials science, and accurate analysis of the nanoparticle topology is essential to determine their properties. Unfortunately, the lack of high-quality annotated datasets drastically hinders the creation of strong segmentation models for nanoscale imaging. To alleviate this problem, we introduce F-ANcGAN, an attention-enhanced cycle consistent generative adversarial system that can be trained using a limited number of data samples and generates realistic scanning electron microscopy (SEM) images directly from segmentation maps. Our model uses a Style U-Net generator and a U-Net segmentation network equipped with self-attention to capture structural relationships and applies augmentation methods to increase the variety of the dataset. The architecture reached a raw FID score of 17.65 for TiO$_2$ dataset generation, with a further reduction in FID score to nearly 10.39 by using efficient post-processing techniques. By facilitating scalable high-fidelity synthetic dataset generation, our approach can improve the effectiveness of downstream segmentation task training, overcoming severe data shortage issues in nanoparticle analysis, thus extending its applications to resource-limited fields.

Problem

• Lack of high-quality annotated nanoparticle datasets hinders segmentation models
• Generating realistic SEM images from limited segmentation maps is challenging
• Data shortage in nanoparticle analysis limits downstream task effectiveness

Innovation

• Attention-enhanced cycle consistent GAN
• Style U-Net generator with self-attention
• Efficient post-processing for FID reduction