Deep learning improves utility of tau PET in the study of Alzheimer's disease.
Ontology highlight
ABSTRACT: Introduction
Positron emission tomography (PET) imaging targeting neurofibrillary tau tangles is increasingly used in the study of Alzheimer's disease (AD), but its utility may be limited by conventional quantitative or qualitative evaluation techniques in earlier disease states. Convolutional neural networks (CNNs) are effective in learning spatial patterns for image classification.Methods
18F-MK6240 (n = 320) and AV-1451 (n = 446) PET images were pooled from multiple studies. We performed iterations with differing permutations of radioligands, heuristics, and architectures. Performance was compared to a standard region of interest (ROI)-based approach on prediction of memory impairment. We visualized attention of the network to illustrate decision making.Results
Overall, models had high accuracy (> 80%) with good average sensitivity and specificity (75% and 82%, respectively), and had comparable or higher accuracy to the ROI standard. Visualizations of model attention highlight known characteristics of tau radioligand binding.Discussion
CNNs could improve tau PET's role in early disease and extend the utility of tau PET across generations of radioligands.
SUBMITTER: Zou J
PROVIDER: S-EPMC8719427 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:A biological research framework to define Alzheimer' disease with dichotomized biomarker measurement was proposed by National Institute on Aging-Alzheimer's Association (NIA-AA). However, it cannot characterize the hierarchy spreading pattern of tau pathology. To reflect in vivo tau progression using biomarker, we constructed a refined topographic 18F-AV-1451 tau PET staging scheme with longitudinal clinical validation. Seven hundred and thirty-four participants with baseline 18F-AV-1451 tau PET (baseline age 73.9 ± 7.7 years, 375 female) were stratified into five stages by a topographic PET staging scheme. Cognitive trajectories and clinical progression were compared across stages with or without further dichotomy of amyloid status, using linear mixed-effect models and Cox proportional hazard models. Significant cognitive decline was first observed in stage 1 when tau levels only increased in transentorhinal regions. Rates of cognitive decline and clinical progression accelerated from stage 2 to stage 3 and stage 4. Higher stages were also associated with greater CSF phosphorylated tau and total tau concentrations from stage 1. Abnormal tau accumulation did not appear with normal β-amyloid in neocortical regions but prompt cognitive decline by interacting with β-amyloid in temporal regions. Highly accumulated tau in temporal regions independently led to cognitive deterioration. Topographic PET staging scheme have potentials in early diagnosis, predicting disease progression, and studying disease mechanism. Characteristic tau spreading pattern in Alzheimer's disease could be illustrated with biomarker measurement under NIA-AA framework. Clinical-neuroimaging-neuropathological studies in other cohorts are needed to validate these findings.
| S-EPMC8449785 | biostudies-literature
Project description:See Hansson and Mormino (doi:10.1093/brain/awy065) for a scientific commentary on this article.Our objective was to compare different whole-brain and region-specific measurements of within-person change on serial tau PET and evaluate its utility for clinical trials. We studied 126 individuals: 59 cognitively unimpaired with normal amyloid, 37 cognitively unimpaired with abnormal amyloid, and 30 cognitively impaired with an amnestic phenotype and abnormal amyloid. All had baseline amyloid PET and two tau PET, MRI, and clinical assessments. We compared the topography across all cortical regions of interest of tau PET accumulation rates and the rates of four different whole-brain or region-specific meta-regions of interest among the three clinical groups. We computed sample size estimates for change in tau PET, cortical volume, and memory/mental status indices for use as outcome measures in clinical trials. The cognitively unimpaired normal amyloid group had no observable tau accumulation throughout the brain. Tau accumulation rates in cognitively unimpaired abnormal amyloid were low [0.006 standardized uptake value ratio (SUVR), 0.5%, per year] but greater than rates in the cognitively unimpaired normal amyloid group in the basal and mid-temporal, retrosplenial, posterior cingulate, and entorhinal regions of interest. Thus, the earliest elevation in accumulation rates was widespread and not confined to the entorhinal cortex. Tau accumulation rates in the cognitively impaired abnormal amyloid group were 0.053 SUVR (3%) per year and greater than rates in cognitively unimpaired abnormal amyloid in all cortical areas except medial temporal. Rates of accumulation in the four meta-regions of interest differed but only slightly from one another. Among all tau PET meta-regions of interest, sample size estimates were smallest for a temporal lobe composite within cognitively unimpaired abnormal amyloid and for the late Alzheimer's disease meta-region of interest within cognitively impaired abnormal amyloid. The ordering of the sample size estimates by outcome measure was MRI < tau PET < cognitive measures. At a group-wise level, observable rates of short-term serial tau accumulation were only seen in the presence of abnormal amyloid. As disease progressed to clinically symptomatic stages (cognitively impaired abnormal amyloid), observable rates of tau accumulation were seen uniformly throughout the brain providing evidence that tau does not accumulate in one area at a time or in start-stop, stepwise sequence. The information captured by rate measures in different meta-regions of interest, even those with little topographic overlap, was similar. The implication is that rate measurements from simple meta-regions of interest, without the need for Braak-like staging, may be sufficient to capture progressive within-person accumulation of pathologic tau. Tau PET SUVR measures should be an efficient outcome measure in disease-modifying clinical trials.
| S-EPMC5917767 | biostudies-literature
Project description:In Alzheimer's disease, the molecular pathogenesis of the extracellular Aβ-amyloid (Aβ) instigation of intracellular tau accumulation is poorly understood. We employed a high-resolution PET scanner, with low detection thresholds, to examine the Aβ-tau association using a convolutional neural network (CNN), and compared results to a standard voxel-wise linear analysis. The full range of Aβ Centiloid values was highly predicted by the tau topography using the CNN (training R2 = 0.86, validation R2 = 0.75, testing R2 = 0.72). Linear models based on tau-SUVR identified widespread positive correlations between tau accumulation and Aβ burden throughout the brain. In contrast, CNN analysis identified focal clusters in the bilateral medial temporal lobes, frontal lobes, precuneus, postcentral gyrus and middle cingulate. At low Aβ levels, information from the middle cingulate, frontal lobe and precuneus regions was more predictive of Aβ burden, while at high Aβ levels, the medial temporal regions were more predictive of Aβ burden. The data-driven CNN approach revealed new associations between tau topography and Aβ burden.
| S-EPMC9427855 | biostudies-literature
Project description:Abnormal tau inclusions are hallmarks of Alzheimer's disease and predictors of clinical decline. Several tau PET tracers are available for neurodegenerative disease research, opening avenues for molecular diagnosis in vivo. However, few have been approved for clinical use. Understanding the neurobiological basis of PET signal validation remains problematic because it requires a large-scale, voxel-to-voxel correlation between PET and (immuno) histological signals. Large dimensionality of whole human brains, tissue deformation impacting co-registration, and computing requirements to process terabytes of information preclude proper validation. We developed a computational pipeline to identify and segment particles of interest in billion-pixel digital pathology images to generate quantitative, 3D density maps. The proposed convolutional neural network for immunohistochemistry samples, IHCNet, is at the pipeline's core. We have successfully processed and immunostained over 500 slides from two whole human brains with three phospho-tau antibodies (AT100, AT8, and MC1), spanning several terabytes of images. Our artificial neural network estimated tau inclusion from brain images, which performs with ROC AUC of 0.87, 0.85, and 0.91 for AT100, AT8, and MC1, respectively. Introspection studies further assessed the ability of our trained model to learn tau-related features. We present an end-to-end pipeline to create terabytes-large 3D tau inclusion density maps co-registered to MRI as a means to facilitate validation of PET tracers.
| S-EPMC8983026 | biostudies-literature
Project description:As the world's population ages, the number of patients with Alzheimer's disease (AD) is predicted to increase rapidly. The presence of neurofibrillary tangles (NFTs), composed of hyperphosphorylated tau protein, is one of the neuropathological hallmarks of AD brain. Since the presence of NFTs is well correlated with neurodegeneration and cognitive decline in AD, imaging of tau using positron emission tomography (PET) and single-photon emission computed tomography (SPECT) is useful for presymptomatic diagnosis and monitoring of the progression of AD. Therefore, novel PET/SPECT probes for the imaging of tau have been developed. More recently, several probes were tested clinically and evaluated for their utility. This paper reviews the current state of research on the development and evaluation of PET/SPECT probes for the imaging of tau in AD brain.
| S-EPMC4386695 | biostudies-other
Project description:Alzheimer's disease (AD) patients are known to have heterogeneous clinical presentation and pathologic patterns. We hypothesize that AD dementia can be categorized into subtypes based on multimodal imaging biomarkers such as magnetic resonance imaging (MRI), tau positron emission tomography (PET), and amyloid PET. We collected 3T MRI, 18F-THK5351 PET, and 18F-flutemetamol (FLUTE) PET data from 83 patients with AD dementia [Clinical Dementia Rating (CDR) ≤1] and 60 normal controls (NC), and applied surface-based analyses to measure cortical thickness, THK5351 standardized uptake value ratio (SUVR) and FLUTE SUVR for each participant. For the patient group, we performed an agglomerative hierarchical clustering analysis using the three multimodal imaging features on the vertices (n = 3 × 79,950). The identified AD subtypes were compared to NC using general linear models adjusting for age, sex, and years of education. We mapped the effect size within significant cortical regions reaching a corrected p-vertex <0.05 (random field theory). Our surface-based multimodal framework has revealed three distinct subtypes among AD patients: medial temporal-dominant subtype (MT, n = 44), parietal-dominant subtype (P, n = 19), and diffuse atrophy subtype (D, n = 20). The topography of cortical atrophy and THK5351 retention differentiates between the three subtypes. In the case of FLUTE, three subtypes did not show distinct topographical differences, although cortical composite retention was significantly higher in the P type than in the MT type. These three subtypes also differed in demographic and clinical features. In conclusion, AD patients may be clustered into three subtypes with distinct topographical features of cortical atrophy and tau deposition, although amyloid deposition may not differ across the subtypes in terms of topography.
| S-EPMC6710378 | biostudies-literature
Project description:INTRODUCTION:Differential patterns of brain atrophy on structural magnetic resonance imaging (MRI) revealed four reproducible subtypes of Alzheimer's disease (AD): (1) "typical", (2) "limbic-predominant", (3) "hippocampal-sparing", and (4) "mild atrophy". We examined the neurobiological characteristics and clinical progression of these atrophy-defined subtypes. METHODS:The four subtypes were replicated using a clustering method on MRI data in 260 amyloid-?-positive patients with mild cognitive impairment or AD dementia, and we subsequently tested whether the subtypes differed on [18 F]flortaucipir (tau) positron emission tomography, white matter hyperintensity burden, and rate of global cognitive decline. RESULTS:Voxel-wise and region-of-interest analyses revealed the greatest neocortical tau load in hippocampal-sparing (frontoparietal-predominant) and typical (temporal-predominant) patients, while limbic-predominant patients showed particularly high entorhinal tau. Typical patients with AD had the most pronounced white matter hyperintensity load, and hippocampal-sparing patients showed the most rapid global cognitive decline. DISCUSSION:Our data suggest that structural MRI can be used to identify biologically and clinically meaningful subtypes of AD.
| S-EPMC7012375 | biostudies-literature
Project description:Accumulation of paired helical filament tau contributes to neurodegeneration in Alzheimer's disease (AD). 18 F-flortaucipir is a positron emission tomography (PET) radioligand sensitive to tau in AD, but its clinical utility will depend in part on its ability to predict cognitive symptoms in diverse dementia phenotypes associated with selective, regional uptake. We examined associations between 18 F-flortaucipir and cognition in 14 mildly-impaired patients (12 with cerebrospinal fluid analytes consistent with AD pathology) who had amnestic (n = 5) and non-amnestic AD syndromes, including posterior cortical atrophy (PCA, n = 5) and logopenic-variant primary progressive aphasia (lvPPA, n = 4). Amnestic AD patients had deficits in memory; lvPPA in language; and both amnestic AD and PCA patients in visuospatial function. Associations with cognition were tested using sparse regression and compared to associations in anatomical regions-of-interest (ROIs). 18 F-flortaucipir uptake was expected to show regionally-specific correlations with each domain. In multivariate analyses, uptake was elevated in neocortical areas specifically associated with amnestic and non-amnestic syndromes. Uptake in left anterior superior temporal gyrus accounted for 67% of the variance in language performance. Uptake in right lingual gyrus predicted 85% of the variance in visuospatial performance. Memory was predicted by uptake in right fusiform gyrus and cuneus as well as a cluster comprising right anterior hippocampus and amygdala; this eigenvector explained 57% of the variance in patients' scores. These results provide converging evidence for associations between 18 F-flortaucipir uptake, tau pathology, and patients' cognitive symptoms.
| S-EPMC5764792 | biostudies-literature
Project description:A next generation of tau PET tracers for the imaging of Alzheimer's disease and other dementias has recently been developed. Whilst the new compounds have now entered clinical studies, there is limited information available to assess their suitability for clinical applications. Head-to-head comparisons are urgently needed to understand differences in the radiotracer binding profiles. We characterized the binding of the tau tracers PI2620, RO948, MK6240 and JNJ067 in human post-mortem brain tissue from a cohort of 25 dementia cases and age-matched controls using quantitative phosphorimaging with tritium-labelled radiotracers in conjunction with phospho-tau specific immunohistochemistry. The four radiotracers depicted tau inclusions composed of paired helical filaments with high specificity, both in cases with Alzheimer's disease and in primary tauopathy cases with concomitant Alzheimer's disease pathology. In contrast, cortical binding to primary tauopathy in cases without paired helical filament tau was found to be within the range of age-matched controls. Off-target binding to monoamine oxidase B has been overcome, as demonstrated by heterologous blocking studies in basal ganglia tissue. The high variability of cortical tracer binding within the Alzheimer's disease group followed the same pattern with each tracer, suggesting that all compounds are suited to differentiate Alzheimer's disease from other dementias.
| S-EPMC8453387 | biostudies-literature
Project description:SEE SARAZIN ET AL DOI101093/BRAIN/AWW041 FOR A SCIENTIFIC COMMENTARY ON THIS ARTICLE: The advent of the positron emission tomography tracer (18)F-AV1451 provides the unique opportunity to visualize the regional distribution of tau pathology in the living human brain. In this study, we tested the hypothesis that tau pathology is closely linked to symptomatology and patterns of glucose hypometabolism in Alzheimer's disease, in contrast to the more diffuse distribution of amyloid-β pathology. We included 20 patients meeting criteria for probable Alzheimer's disease dementia or mild cognitive impairment due to Alzheimer's disease, presenting with a variety of clinical phenotypes, and 15 amyloid-β-negative cognitively normal individuals, who underwent (18)F-AV1451 (tau), (11)C-PiB (amyloid-β) and (18)F-FDG (glucose metabolism) positron emission tomography, apolipoprotein E (APOE) genotyping and neuropsychological testing. Voxel-wise contrasts against controls (at P < 0.05 family-wise error corrected) showed that (18)F-AV1451 and (18)F-FDG patterns in patients with posterior cortical atrophy ('visual variant of Alzheimer's disease', n = 7) specifically targeted the clinically affected posterior brain regions, while (11)C-PiB bound diffusely throughout the neocortex. Patients with an amnestic-predominant presentation (n = 5) showed highest (18)F-AV1451 retention in medial temporal and lateral temporoparietal regions. Patients with logopenic variant primary progressive aphasia ('language variant of Alzheimer's disease', n = 5) demonstrated asymmetric left greater than right hemisphere (18)F-AV1451 uptake in three of five patients. Across 30 FreeSurfer-defined regions of interest in 16 Alzheimer's disease patients with all three positron emission tomography scans available, there was a strong negative association between (18)F-AV1451 and (18)F-FDG uptake (Pearson's r = -0.49 ± 0.07, P < 0.001) and less pronounced positive associations between (11)C-PiB and (18)F-FDG (Pearson's r = 0.16 ± 0.09, P < 0.001) and (18)F-AV1451 and (11)C-PiB (Pearson's r = 0.18 ± 0.09, P < 0.001). Voxel-wise linear regressions thresholded at P < 0.05 (uncorrected) showed that, across all patients, younger age was associated with greater (18)F-AV1451 uptake in wide regions of the neocortex, while older age was associated with increased (18)F-AV1451 in the medial temporal lobe. APOE ϵ4 carriers showed greater temporal and parietal (18)F-AV1451 uptake than non-carriers. Finally, worse performance on domain-specific neuropsychological tests was associated with greater (18)F-AV1451 uptake in key regions implicated in memory (medial temporal lobes), visuospatial function (occipital, right temporoparietal cortex) and language (left > right temporoparietal cortex). In conclusion, tau imaging-contrary to amyloid-β imaging-shows a strong regional association with clinical and anatomical heterogeneity in Alzheimer's disease. Although preliminary, these results are consistent with and expand upon findings from post-mortem, animal and cerebrospinal fluid studies, and suggest that the pathological aggregation of tau is closely linked to patterns of neurodegeneration and clinical manifestations of Alzheimer's disease.
| S-EPMC5006248 | biostudies-literature