AI-Augmented Thyroid Scintigraphy for Robust Classification

2025-03-01

AI Summary

To address poor generalizability of deep learning models on thyroid scintigraphy images—caused by limited sample size and severe class imbalance—this paper systematically evaluates multiple data augmentation strategies for ResNet18-based classification. We innovatively introduce Flow Matching (FM), a generative augmentation method, to nuclear medicine imaging of the thyroid for the first time, proposing an integrated paradigm: “Original data + Flow Matching + Conventional Augmentation” (O+FM+CA). Experimental results demonstrate that O+FM+CA achieves the highest classification accuracy across four diagnostic categories—diffuse goiter, nodular goiter, normal thyroid, and thyroiditis. Wilcoxon signed-rank tests confirm its statistically significant superiority over alternative generative methods, including Stable Diffusion. Moreover, the approach substantially improves cross-disease classification fairness, mitigates dataset bias, and enhances clinical applicability.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Computer Vision: Diffusion Models for Vision
• Natural Language Processing: Generation

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Responsible Web: Data and user privacy-enhancing technologies for the Web

Abstract

Thyroid scintigraphy is a key imaging modality for diagnosing thyroid disorders. Deep learning models for thyroid scintigraphy classification often face challenges due to limited and imbalanced datasets, leading to suboptimal generalization. In this study, we investigate the effectiveness of different data augmentation techniques including Stable Diffusion (SD), Flow Matching (FM), and Conventional Augmentation (CA) to enhance the performance of a ResNet18 classifier for thyroid condition classification. Our results showed that FM-based augmentation consistently outperforms SD-based approaches, particularly when combined with original (O) data and CA (O+FM+CA), achieving both high accuracy and fair classification across Diffuse Goiter (DG), Nodular Goiter (NG), Normal (NL), and Thyroiditis (TI) cases. The Wilcoxon statistical analysis further validated the superiority of O+FM and its variants (O+FM+CA) over SD-based augmentations in most scenarios. These findings highlight the potential of FM-based augmentation as a superior approach for generating high-quality synthetic thyroid scintigraphy images and improving model generalization in medical image classification.

Problem

• Enhance thyroid scintigraphy classification using data augmentation.
• Address limited, imbalanced datasets in deep learning models.
• Improve model generalization for thyroid disorder diagnosis.

Innovation

• Flow Matching outperforms Stable Diffusion in augmentation
• Combining original data with Flow Matching enhances accuracy
• Flow Matching improves generalization in medical image classification