Safeguarding Generative AI Applications in Preclinical Imaging through Hybrid Anomaly Detection

2025-08-11

AI Summary

Generative AI systems in preclinical nuclear medicine imaging—such as Pose2Xray and DosimetrEYE—are prone to anomalous image synthesis due to data bias or model mismatch, posing risks to quantitative accuracy and experimental reproducibility. Method: This paper proposes a hybrid anomaly detection framework tailored for multimodal biomedical image generation. It integrates autoencoder-based reconstruction error analysis, feature-space density clustering, and statistical significance testing to enable real-time, interpretable quality monitoring during X-ray image and 3D dosimetric map generation. Contribution/Results: Deployed on an industrial-grade AI platform, the framework achieves end-to-end automated regulatory oversight, attaining ≥97% anomaly detection rate with zero critical false negatives in production. Compared to manual review, it significantly reduces quality control overhead while enabling safe, high-throughput scaling of preclinical studies.

Technology Category

Machine Learning: Calibration & Uncertainty QuantificationComputer Vision: Multi-modal VisionData Mining & Knowledge Management: Anomaly/Outlier Detection

Application Category

Social Networks and Social Media: Generative AI / large language models and their impact on social systemsResponsible Web: Machine-in-the-loop, human agency and autonomySearch and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

Generative AI holds great potentials to automate and enhance data synthesis in nuclear medicine. However, the high-stakes nature of biomedical imaging necessitates robust mechanisms to detect and manage unexpected or erroneous model behavior. We introduce development and implementation of a hybrid anomaly detection framework to safeguard GenAI models in BIOEMTECH's eyes(TM) systems. Two applications are demonstrated: Pose2Xray, which generates synthetic X-rays from photographic mouse images, and DosimetrEYE, which estimates 3D radiation dose maps from 2D SPECT/CT scans. In both cases, our outlier detection (OD) enhances reliability, reduces manual oversight, and supports real-time quality control. This approach strengthens the industrial viability of GenAI in preclinical settings by increasing robustness, scalability, and regulatory compliance.

Problem

• Detect and manage unexpected GenAI behavior in biomedical imaging
• Enhance reliability of synthetic X-ray and radiation dose generation
• Improve robustness and compliance of GenAI in preclinical applications

Innovation

• Hybrid anomaly detection for GenAI models
• Outlier detection enhances reliability and quality
• Real-time quality control in preclinical imaging