TauAD: MRI-free Tau Anomaly Detection in PET Imaging via Conditioned Diffusion Models

Lujia Zhong,Shuo Huang,Jiaxin Yue,Jianwei Zhang,Zhiwei Deng,Wenhao Chi,Yonggang Shi

2024-05-22

Abstract:The emergence of tau PET imaging over the last decade has enabled Alzheimer's disease (AD) researchers to examine tau pathology in vivo and more effectively characterize the disease trajectories of AD. Current tau PET analysis methods, however, typically perform inferences on large cortical ROIs and are limited in the detection of localized tau pathology that varies across subjects. Furthermore, a high-resolution MRI is required to carry out conventional tau PET analysis, which is not commonly acquired in clinical practices and may not be acquired for many elderly patients with dementia due to strong motion artifacts, claustrophobia, or certain metal implants. In this work, we propose a novel conditional diffusion model to perform MRI-free anomaly detection from tau PET imaging data. By including individualized conditions and two complementary loss maps from pseudo-healthy and pseudo-unhealthy reconstructions, our model computes an anomaly map across the entire brain area that allows simply training a support vector machine (SVM) for classifying disease severity. We train our model on ADNI subjects (n=534) and evaluate its performance on a separate dataset from the preclinical subjects of the A4 clinical trial (n=447). We demonstrate that our method outperforms baseline generative models and the conventional Z-score-based method in anomaly localization without mis-detecting off-target bindings in sub-cortical and out-of-brain areas. By classifying the A4 subjects according to their anomaly map using the SVM trained on ADNI data, we show that our method can successfully group preclinical subjects with significantly different cognitive functions, which further demonstrates the effectiveness of our method in capturing biologically relevant anomaly in tau PET imaging.

Image and Video Processing,Computer Vision and Pattern Recognition

What problem does this paper attempt to address?

This paper proposes a solution to the problem of detecting tau abnormalities in positron emission tomography (PET) imaging without the need for magnetic resonance imaging (MRI). Current methods often rely on inference from large cortical regions and are limited in detecting local tau lesions that vary across individuals. In addition, conventional tau PET analysis requires high-resolution MRI, which is not commonly available in clinical practice, especially for elderly individuals with movement disorders, claustrophobia, or metal implants. Therefore, the paper proposes a conditional diffusion model that calculates abnormal maps of the entire brain region from PET imaging data using personalized conditions and two complementary loss maps reconstructed from pseudo-healthy and pseudo-unhealthy individuals, thereby simplifying the classification training of disease severity by support vector machines (SVM). Experimental results show that this method outperforms baseline generative models and traditional Z-score methods in anomaly localization, without falsely detecting non-target structures in the subcortex and extracranial regions. By classifying subjects from the A4 study according to the SVM trained on the ADNI dataset, it is demonstrated that this method can successfully group clinically preclinical individuals with significantly different cognitive functions, demonstrating its efficacy in capturing biologically relevant abnormalities in PET imaging.