Enhancing Fetal Plane Classification Accuracy with Data Augmentation Using Diffusion Models

2025-01-25

AI Summary

To address the limited availability of annotated fetal ultrasound images, which constrains classification model performance, this work pioneers the systematic application of diffusion models to this domain, proposing a novel two-stage transfer learning paradigm: “synthetic-data-first pretraining followed by fine-tuning on real data.” Methodologically, it integrates diffusion-based image generation, supervised classification architectures (ResNet/ViT), multi-scale data augmentation, and domain alignment techniques. Evaluated across multiple clinical fetal ultrasound datasets, the approach achieves an average classification accuracy improvement of 5.2%; under few-shot settings (≤50 samples per class), the gain reaches 11.7%, significantly outperforming GAN-based and interpolation-based augmentation baselines. This work establishes a generalizable, synthesis–transfer co-design framework for medical image analysis under data-scarce conditions.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language

Application Category

• Web Mining and Content Analysis: Large pretrained models with web data
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Ultrasound imaging is widely used in medical diagnosis, especially for fetal health assessment. However, the availability of high-quality annotated ultrasound images is limited, which restricts the training of machine learning models. In this paper, we investigate the use of diffusion models to generate synthetic ultrasound images to improve the performance on fetal plane classification. We train different classifiers first on synthetic images and then fine-tune them with real images. Extensive experimental results demonstrate that incorporating generated images into training pipelines leads to better classification accuracy than training with real images alone. The findings suggest that generating synthetic data using diffusion models can be a valuable tool in overcoming the challenges of data scarcity in ultrasound medical imaging.

Problem

• Ultrasound Image
• Data Scarcity
• Fetal Part Recognition

Innovation

• Diffusion Model
• Ultrasound Image Synthesis
• Fetal Plane Recognition