Dataset Distillation in Medical Imaging: A Feasibility Study

2024-07-19

arXiv.org

AI Summary

Medical imaging data sharing is hindered by privacy concerns and high transmission costs. This work systematically investigates the feasibility and efficacy of data distillation in this domain. We conduct the first comprehensive comparative evaluation of mainstream distillation methods—including Dataset Condensation (DC), Distribution Matching (DM), and Dataset Synthesis via Augmentation (DSA)—on multi-source medical imaging data, incorporating medical-specific preprocessing and cross-dataset generalization validation. We propose a novel representativeness metric for distilled samples that predicts downstream model performance, and empirically demonstrate that tiny synthetic datasets (1–5% of original size) can recover ≥90% of full-data training performance. Experiments confirm that distilled data suffices for training diagnostic-grade models while preserving privacy and substantially improving cross-institutional collaboration efficiency. Our core contributions are: (1) establishing the applicability boundaries of medical image data distillation; (2) introducing a performance-prediction paradigm; and (3) providing practical, benchmarked guidelines for real-world deployment.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Data Mining & Knowledge Management: Data Compression

Application Category

• Security and Privacy: Data transparency and provenance
• Web Mining and Content Analysis: Large pretrained models with web data
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

Data sharing in the medical image analysis field has potential yet remains underappreciated. The aim is often to share datasets efficiently with other sites to train models effectively. One possible solution is to avoid transferring the entire dataset while still achieving similar model performance. Recent progress in data distillation within computer science offers promising prospects for sharing medical data efficiently without significantly compromising model effectiveness. However, it remains uncertain whether these methods would be applicable to medical imaging, since medical and natural images are distinct fields. Moreover, it is intriguing to consider what level of performance could be achieved with these methods. To answer these questions, we conduct investigations on a variety of leading data distillation methods, in different contexts of medical imaging. We evaluate the feasibility of these methods with extensive experiments in two aspects: 1) Assess the impact of data distillation across multiple datasets characterized by minor or great variations. 2) Explore the indicator to predict the distillation performance. Our extensive experiments across multiple medical datasets reveal that data distillation can significantly reduce dataset size while maintaining comparable model performance to that achieved with the full dataset, suggesting that a small, representative sample of images can serve as a reliable indicator of distillation success. This study demonstrates that data distillation is a viable method for efficient and secure medical data sharing, with the potential to facilitate enhanced collaborative research and clinical applications.

Problem

• Evaluate data distillation in medical imaging feasibility
• Assess impact across varied medical datasets
• Explore indicators for predicting distillation performance

Innovation

• Dataset distillation reduces size
• Maintains comparable model performance
• Facilitates secure medical data sharing