Dual-Mode Deep Anomaly Detection for Medical Manufacturing: Structural Similarity and Feature Distance

2025-09-06

AI Summary

Medical device manufacturing faces challenges including severe scarcity of defective samples, high-resolution imagery, significant data distribution shift across production batches, and stringent regulatory compliance requirements. Method: We propose a dual-mode unsupervised deep anomaly detection framework integrating an attention-guided autoencoder with a dual-criterion scoring mechanism: (i) a 4-scale Multi-Scale Structural Similarity (4-MS-SSIM) metric for pixel-level reconstruction fidelity, and (ii) Mahalanobis distance in the latent feature space to enhance sensitivity to distribution shifts. The framework supports both real-time online inspection and post-production regulatory auditing. Thresholds are adaptively determined via dimensionality-reduced latent feature analysis, enabling hybrid unsupervised/supervised calibration. Results: Evaluated on an industrial dataset using only 10% defective samples, the two modes achieve accuracy scores of 0.931 and 0.722, respectively—substantially outperforming baseline methods—and demonstrate practical viability and regulatory robustness for high-stakes AI deployment.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Computer Vision: Multi-modal Vision
• Data Mining & Knowledge Management: Anomaly/Outlier Detection

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

Automating visual inspection in medical device manufacturing remains challenging due to small and imbalanced datasets, high-resolution imagery, and stringent regulatory requirements. This work proposes two attention-guided autoencoder architectures for deep anomaly detection designed to address these constraints. The first employs a structural similarity-based anomaly score (4-MS-SSIM), offering lightweight and accurate real-time defect detection, yielding ACC 0.903 (unsupervised thresholding) and 0.931 (supervised thresholding) on the - Surface Seal Image - Test split with only 10% of defective samples. The second applies a feature-distance approach using Mahalanobis scoring on reduced latent features, providing high sensitivity to distributional shifts for supervisory monitoring, achieving ACC 0.722 with supervised thresholding. Together, these methods deliver complementary capabilities: the first supports reliable inline inspection, while the second enables scalable post-production surveillance and regulatory compliance monitoring. Experimental results demonstrate that both approaches surpass re-implemented baselines and provide a practical pathway for deploying deep anomaly detection in regulated manufacturing environments, aligning accuracy, efficiency, and the regulatory obligations defined for high-risk AI systems under the EU AI Act.

Problem

• Automating visual inspection in medical device manufacturing
• Addressing small and imbalanced datasets with high-resolution imagery
• Meeting stringent regulatory requirements for anomaly detection

Innovation

• Attention-guided autoencoder architectures for anomaly detection
• Structural similarity-based scoring for real-time defect detection
• Feature-distance approach using Mahalanobis for distributional shifts