Improving Quality Control Of MRI Images Using Synthetic Motion Data

2025-01-31

AI Summary

Addressing the challenges of scarce ground-truth annotations, severe class imbalance, and high subjectivity in MRI quality assessment, this paper proposes a pretraining and transfer learning framework leveraging controllably synthesized motion artifacts. Our method innovatively employs physics-informed synthetic artifact data—generated via forward MRI acquisition modeling—for self-supervised pretraining, thereby substantially enhancing robustness in low-quality image detection under limited labeled data and markedly reducing reliance on manual annotation. The framework integrates MRI artifact modeling, deep transfer learning, and a lightweight CNN architecture. Evaluated on multicenter real-world MRI datasets, it achieves a 12.6% improvement in quality classification accuracy, reduces training time by 40%, and significantly lowers computational resource requirements. This enables scalable, cost-effective deployment of automated quality control in resource-constrained clinical settings.

Technology Category

• Computer Vision: Motion & Tracking
• Machine Learning: Adversarial Learning & Robustness
• Knowledge Representation and Reasoning: Qualitative Reasoning

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

MRI quality control (QC) is challenging due to unbalanced and limited datasets, as well as subjective scoring, which hinder the development of reliable automated QC systems. To address these issues, we introduce an approach that pretrains a model on synthetically generated motion artifacts before applying transfer learning for QC classification. This method not only improves the accuracy in identifying poor-quality scans but also reduces training time and resource requirements compared to training from scratch. By leveraging synthetic data, we provide a more robust and resource-efficient solution for QC automation in MRI, paving the way for broader adoption in diverse research settings.

Problem

• MRI Quality Control
• Data Insufficiency
• Subjectivity in Evaluation

Innovation

• Synthetic Motion Data
• MRI Image Quality Assessment
• Automated Quality Control