Weakly-Supervised PET Anomaly Detection using Implicitly-Guided Attention-Conditional Counterfactual Diffusion Modeling: a Multi-Center, Multi-Cancer, and Multi-Tracer Study

2024-04-30

AI Summary

To address the high cost of pixel-level annotations in PET lesion detection, this paper proposes a weakly supervised anomaly detection method requiring only image-level healthy/abnormal labels—no lesion localization annotations. Our approach leverages an implicitly guided, attention-conditioned counterfactual diffusion model to synthesize “abnormal→healthy” PET images; lesions are localized via differential maps between original and synthesized images. We present the first systematic evaluation of weakly supervised generalization across multi-center, multi-cancer-type, and multi-radiotracer PET data (2,652 cases), and uncover the critical role of attention mechanisms in detecting subtle lesions. Compared to GAN- and VAE-based baselines and conventional SUVmax thresholding, our method achieves a 12.7% improvement in sensitivity, significantly outperforming existing approaches. The implementation is publicly available.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Data Mining & Knowledge Management: Anomaly/Outlier Detection
• Machine Learning: Unsupervised & Self-Supervised Learning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

Minimizing the need for pixel-level annotated data to train PET lesion detection and segmentation networks is highly desired and can be transformative, given time and cost constraints associated with expert annotations. Current un-/weakly-supervised anomaly detection methods rely on autoencoder or generative adversarial networks trained only on healthy data; however GAN-based networks are more challenging to train due to issues with simultaneous optimization of two competing networks, mode collapse, etc. In this paper, we present the weakly-supervised Implicitly guided COuNterfactual diffusion model for Detecting Anomalies in PET images (IgCONDA-PET). The solution is developed and validated using PET scans from six retrospective cohorts consisting of a total of 2652 cases containing both local and public datasets. The training is conditioned on image class labels (healthy vs. unhealthy) via attention modules, and we employ implicit diffusion guidance. We perform counterfactual generation which facilitates"unhealthy-to-healthy"domain translation by generating a synthetic, healthy version of an unhealthy input image, enabling the detection of anomalies through the calculated differences. The performance of our method was compared against several other deep learning based weakly-supervised or unsupervised methods as well as traditional methods like 41% SUVmax thresholding. We also highlight the importance of incorporating attention modules in our network for the detection of small anomalies. The code is publicly available at: https://github.com/ahxmeds/IgCONDA-PET.git.

Problem

• Detect PET anomalies with minimal annotated data
• Improve weakly-supervised anomaly detection using diffusion models
• Generate healthy counterfactuals for anomaly identification

Innovation

• Weakly-supervised anomaly detection
• Attention-guided counterfactual diffusion
• Multi-center PET image analysis