Principal Components Analysis of Brain Metabolism Predicts Development of Alzheimer Dementia

Using interpretable deep learning radiomics model to diagnose and predict progression of early AD disease spectrum: a preliminary [18F]FDG PET study

OBJECTIVES

In this study, we propose an interpretable deep learning radiomics (IDLR) model based on [18F]FDG PET images to diagnose the clinical spectrum of Alzheimer's disease (AD) and predict the progression from mild cognitive impairment (MCI) to AD.

METHODS

This multicentre study included 1962 subjects from two ethnically diverse, independent cohorts (a Caucasian cohort from ADNI and an Asian cohort merged from two hospitals in China). The IDLR model involved feature extraction, feature selection, and classification/prediction. We evaluated the IDLR model's ability to distinguish between subjects with different cognitive statuses and MCI trajectories (sMCI and pMCI) and compared results with radiomic and deep learning (DL) models. A Cox model tested the IDLR signature's predictive capability for MCI to AD progression. Correlation analyses identified critical IDLR features and verified their clinical diagnostic value.

RESULTS

The IDLR model achieved the best classification results for subjects with different cognitive statuses as well as in those with MCI with distinct trajectories, with an accuracy of 76.51% [72.88%, 79.60%], (95% confidence interval, CI) while those of radiomic and DL models were 69.13% [66.28%, 73.12%] and 73.89% [68.99%, 77.89%], respectively. According to the Cox model, the hazard ratio (HR) of the IDLR model was 1.465 (95% CI: 1.236-1.737, p < 0.001). Moreover, three crucial IDLR features were significantly different across cognitive stages and were significantly correlated with cognitive scale scores (p < 0.01).

CONCLUSIONS

Preliminary results demonstrated that the IDLR model based on [18F]FDG PET images enhanced accuracy in diagnosing the clinical spectrum of AD.

KEY POINTS

Question The study addresses the lack of interpretability in existing DL classification models for diagnosing the AD spectrum. Findings The proposed interpretable DL radiomics model, using radiomics-supervised DL features, enhances interpretability from traditional DL models and improves classification accuracy. Clinical relevance The IDLR model interprets DL features through radiomics supervision, potentially advancing the application of DL in clinical classification tasks.

Phenotype-specific metabolic patterns in Posterior cortical atrophy and early-onset typical Alzheimer's disease

OBJECTIVE

Posterior cortical atrophy (PCA) is generally considered an atypical variant of Alzheimer's disease (AD) and is an important component of early-onset AD. Symptomatologic heterogeneity has led to a high rate of misdiagnosis or delayed diagnosis of early-onset AD. We sought to establish the phenotypic-specific metabolic patterns of PCA and early-onset typical AD (tAD) and to assess whether phenotype-specific neuroimaging biomarkers are more valuable for disease recognition.

METHODS

Patients accepting 18F-FDG PET with an onset age younger than 65 years (PCA, n = 40; early-onset tAD, n = 37; behavioral variant frontotemporal dementia (bv-FTD), n = 35) and healthy controls (HCs, n = 30) were enrolled and divided into two cohorts for pattern establishment and validation, respectively. Similarities and differences between patterns were assessed by pattern topography, expression, classification performance and correlation with clinical severity.

RESULTS

PCA-related pattern (PCARP) was characterized by extensively relative hypometabolism in the parietal lobe, occipital lobe, temporal lobe, cingulate gyrus, and relative hypermetabolism mainly in vermis, thalamus. Early-onset tAD-related pattern (EOtADRP) was characterized by relative hypometabolism mainly in the middle frontal gyrus, angular gyrus, precuneus, middle temporal gyrus, cingulate gyrus, caudate, and relative hypermetabolism mainly in vermis, thalamus, postcentral gyrus. PCARP and EOtADRP were closely related in topography (r = 0.909, P < 0.001) and expression (r = 0.862, P < 0.001). High accuracies in distinguishing corresponding patient group from HC were found in both, while only PCARP was capable of phenotype discrimination (PCA versus early-onset tAD; area under the receiver operating characteristic curve [AUC] = 0.84-0.88 for PCARP, AUC = 0.57-0.62 for EOtADRP) and distinguishment between PCA/early-onset tAD and bv-FTD (AUC = 1.00/0.91 for PCARP, AUC = 0.73/0.62 for EOtADRP). PCARP showed great potential in detecting clinical severity in both phenotypes whereas EOtADRP only worked in early-onset tAD.

CONCLUSION

PCARP outperformed EOtADRP in phenotype discrimination with better potential in severity assessment.

Neuropathological correlates of vulnerability and resilience in the cerebellum in Alzheimer's disease

INTRODUCTION

We investigated whether the cerebellum develops neuropathology that correlates with well-accepted Alzheimer's disease (AD) neuropathological markers and cognitive status.

METHODS

We studied cerebellar cytoarchitecture in a cohort (N = 30) of brain donors. In a larger cohort (N = 605), we queried whether the weight of the contents of the posterior fossa (PF), which contains primarily cerebellum, correlated with dementia status.

RESULTS

Although there was no granular layer (GL) cell loss, GL area was lower in AD cases, particularly in the lateral cerebellum. Lower numbers of mossy fiber synaptic terminals in the cerebellar GL of AD cases correlated with Braak stages IV-VI. PF content weight correlated with dementia independently of age, neuropathology, and education. In addition, we found that a measure of the relative size of the PF content weight to total brain weight correlated with less dementia.

DISCUSSION

These results confirm that the cerebellum is not spared neuropathological damage in AD.

HIGHLIGHTS

Novel evidence of cerebellar atrophy in the granule cell layer of the lateral cerebellar cortex (or 'cognitive cerebellum'), and loss of a specific cerebellar synapse type in this region, the cerebellar glomerulus. Both correlated with dementia status and Braak stages IV through VI, in a cohort with complete neuropathological characterization. Although there have been recent brain imaging studies suggesting a role for cerebellum in Alzheimer's disease, we believe our study constitutes some of the most concrete neuropathological evidence to date of anatomic and synaptic substrates that are disrupted in AD. These changes in this cerebellar region may even play a role in the etiology of cognitive symptoms. Novel evidence that individuals with lower postmortem cerebellar weights showed more cognitive decline, independent of classical neuropathology markers such as Braak stage, Thal phase, or Corsortium to Establish a Registry for Alzheimer's Disease (CERAD) score, suggesting a role for this brain region in dementia, using advanced statistical analysis of a large unbiased population cohort (n = 605), the Adult Changes in Thought (ACT) study. Conversely, a measure of how intact the cerebellum was correlated with less dementia, independent of classical neuropathology markers and cerebral cortical weight, again, in the ACT cohort of 605 brain donors. We believe that this novel finding has relevance and implications for the identification of resilience factors, which may protect against the development of dementia.

Exploratory Analysis of Sleep Deprivation Effects on Gene Expression and Regional Brain Metabolism

Introduction

Sleep deprivation affects cognitive performance and immune function, yet its mechanisms and biomarkers remain unclear. This study explored the relationships among gene expression, brain metabolism, sleep deprivation, and sex differences.

Methods

Fluorodeoxyglucose-18 positron emission tomography measured brain metabolism in regions of interest, and RNA analysis of blood samples assessed gene expression pre- and post-sleep deprivation. Mixed model regression and principal component analysis identified significant genes and regional metabolic changes.

Results

There were 23 and 28 differentially expressed probe sets for the main effects of sex and sleep deprivation, respectively, and 55 probe sets for their interaction (FDR-corrected p < 0.05). Functional analysis of genes affected by sleep deprivation revealed pathway enrichment in nucleoplasm- and UBL conjugation-related genes. Genes with significant sex effects mapped to chromosomes Y and 19 (Benjamini-Hochberg FDR p < 0.05), with 11 genes (4%) and 29 genes (10.5%) involved, respectively. Differential gene expression highlighted sex-based differences in innate and adaptive immunity. For brain metabolism, sleep deprivation resulted in significant decreases in the left insula, left medial prefrontal cortex (BA32), left somatosensory cortex (BA1/2), and left motor premotor cortex (BA6) and increases in the right inferior longitudinal fasciculus, right primary visual cortex (BA17), right amygdala, left cerebellum, and bilateral pons.

Conclusion

Sleep deprivation broadly impacts brain metabolism, gene expression, and immune function, revealing cellular stress responses and hemispheric vulnerability. These findings enhance our understanding of the molecular and functional effects of sleep deprivation.

Cerebellum in neurodegenerative diseases: Advances, challenges, and prospects

Neurodegenerative diseases (NDs) are a group of neurological disorders characterized by the progressive dysfunction of neurons and glial cells, leading to their structural and functional degradation in the central and/or peripheral nervous system. Historically, research on NDs has primarily focused on the brain, brain stem, or spinal cord associated with disease-related symptoms, often overlooking the role of the cerebellum. However, an increasing body of clinical and biological evidence suggests a significant connection between the cerebellum and NDs. In several NDs, cerebellar pathology and biochemical changes may start in the early disease stages. This article provides a comprehensive update on the involvement of the cerebellum in the clinical features and pathogenesis of multiple NDs, suggesting that the cerebellum is involved in the onset and progression of NDs through various mechanisms, including specific neurodegeneration, neuroinflammation, abnormal mitochondrial function, and altered metabolism. Additionally, this review highlights the significant therapeutic potential of cerebellum-related treatments for NDs.

Cerebral Glucose Metabolism Is a Valuable Predictor of Survival in Patients with Lewy Body Diseases

OBJECTIVE

Patients with Lewy body diseases have an increased risk of dementia, which is a significant predictor for survival. Posterior cortical hypometabolism on [18F]fluorodeoxyglucose positron emission tomography (PET) precedes the development of dementia by years. We therefore examined the prognostic value of cerebral glucose metabolism for survival.

METHODS

We enrolled patients diagnosed with Parkinson's disease (PD), Parkinson's disease with dementia, or dementia with Lewy bodies who underwent [18F]fluorodeoxyglucose PET. Regional cerebral metabolism of each patient was analyzed by determining the expression of the PD-related cognitive pattern (Z-score) and by visual PET rating. We analyzed the predictive value of PET for overall survival using Cox regression analyses (age- and sex-corrected) and calculated prognostic indices for the best model.

RESULTS

Glucose metabolism was a significant predictor of survival in 259 included patients (n = 118 events; hazard ratio: 1.4 [1.2-1.6] per Z-score; hazard ratio: 1.8 [1.5-2.2] per visual PET rating score; both p < 0.0001). Risk stratification with visual PET rating scores yielded a median survival of 4.8, 6.8, and 12.9 years for patients with severe, moderate, and mild posterior cortical hypometabolism (median survival not reached for normal cortical metabolism). Stratification into 5 groups based on the prognostic index revealed 10-year survival rates of 94.1%, 78.3%, 34.7%, 0.0%, and 0.0%.

INTERPRETATION

Regional cerebral glucose metabolism is a significant predictor of survival in Lewy body diseases and may allow an earlier survival prediction than the clinical milestone "dementia." Thus, [18F]fluorodeoxyglucose PET may improve the basis for therapy decisions, especially for invasive therapeutic procedures like deep brain stimulation in Parkinson's disease. ANN NEUROL 2024;96:539-550.

PCA and logistic regression in 2-[18F]FDG PET neuroimaging as an interpretable and diagnostic tool for Alzheimer's disease

Objective.to develop an optimization and training pipeline for a classification model based on principal component analysis and logistic regression using neuroimages from PET with 2-[18F]fluoro-2-deoxy-D-glucose (FDG PET) for the diagnosis of Alzheimer's disease (AD).Approach.as training data, 200 FDG PET neuroimages were used, 100 from the group of patients with AD and 100 from the group of cognitively normal subjects (CN), downloaded from the repository of the Alzheimer's Disease Neuroimaging Initiative (ADNI). Regularization methods L1 and L2 were tested and their respective strength varied by the hyperparameter C. Once the best combination of hyperparameters was determined, it was used to train the final classification model, which was then applied to test data, consisting of 192 FDG PET neuroimages, 100 from subjects with no evidence of AD (nAD) and 92 from the AD group, obtained at the Centro de Diagnóstico por Imagem (CDI).Main results.the best combination of hyperparameters was L1 regularization andC≈ 0.316. The final results on test data were accuracy = 88.54%, recall = 90.22%, precision = 86.46% and AUC = 94.75%, indicating that there was a good generalization to neuroimages outside the training set. Adjusting each principal component by its respective weight, an interpretable image was obtained that represents the regions of greater or lesser probability for AD given high voxel intensities. The resulting image matches what is expected by the pathophysiology of AD.Significance.our classification model was trained on publicly available and robust data and tested, with good results, on clinical routine data. Our study shows that it serves as a powerful and interpretable tool capable of assisting in the diagnosis of AD in the possession of FDG PET neuroimages. The relationship between classification model output scores and AD progression can and should be explored in future studies.

The Alzheimer's Disease Neuroimaging Initiative in the era of Alzheimer's disease treatment: A review of ADNI studies from 2021 to 2022

The Alzheimer's Disease Neuroimaging Initiative (ADNI) aims to improve Alzheimer's disease (AD) clinical trials. Since 2006, ADNI has shared clinical, neuroimaging, and cognitive data, and biofluid samples. We used conventional search methods to identify 1459 publications from 2021 to 2022 using ADNI data/samples and reviewed 291 impactful studies. This review details how ADNI studies improved disease progression understanding and clinical trial efficiency. Advances in subject selection, detection of treatment effects, harmonization, and modeling improved clinical trials and plasma biomarkers like phosphorylated tau showed promise for clinical use. Biomarkers of amyloid beta, tau, neurodegeneration, inflammation, and others were prognostic with individualized prediction algorithms available online. Studies supported the amyloid cascade, emphasized the importance of neuroinflammation, and detailed widespread heterogeneity in disease, linked to genetic and vascular risk, co-pathologies, sex, and resilience. Biological subtypes were consistently observed. Generalizability of ADNI results is limited by lack of cohort diversity, an issue ADNI-4 aims to address by enrolling a diverse cohort.

The Relationship between Alzheimer's Disease and Ferroptosis: A Bibliometric Study Based on Citespace

BACKGROUND

The potential relationship between Alzheimer's Disease (AD) and ferroptosis has received considerable attention, yet there is no comprehensive visualization analysis in this field. This study aimed to explore the research frontiers and hotspots through bibliometric analysis.

METHODS

Literature related to AD and ferroptosis was collected from the Web of Science Core Collection. Data, including countries, authors, institutions, journals, and keywords, were analyzed by Tableau Public Desktop and Citespace software.

RESULTS

A total of 305 articles published between January 1st, 2013, and December 31st, 2023, were included, and the number of articles on the relationship between AD and ferroptosis has increased annually, with the largest number reported from China (162 articles). The articles from Professor SJ Dixon were cited most frequently. Among the top ten most cited articles, four were published in top journals. The University of Melbourne emerged as the institution with the highest number of publications (27 articles). Among the journals, most of the articles were published in Frontiers in Aging Neuroscience (13 articles, accounting for 4.26%). The co-occurrence analysis of keywords revealed that major hotspots in this field contained oxidative stress, cell death, and lipid peroxidation. Keyword burst analysis indicated that antioxidant was the term with the longest duration of high interest, while clustering analysis showed that this research area primarily focused on amyloid precursor protein, drug development, and diagnostic models.

CONCLUSION

Bibliometric analyses were conducted to comprehensively present the research progress and trends on the relationship between AD and ferroptosis, providing valuable evidence for future research in related fields.

Artificial intelligence for diagnostic and prognostic neuroimaging in dementia: A systematic review

INTRODUCTION

Artificial intelligence (AI) and neuroimaging offer new opportunities for diagnosis and prognosis of dementia.

METHODS

We systematically reviewed studies reporting AI for neuroimaging in diagnosis and/or prognosis of cognitive neurodegenerative diseases.

RESULTS

A total of 255 studies were identified. Most studies relied on the Alzheimer's Disease Neuroimaging Initiative dataset. Algorithmic classifiers were the most commonly used AI method (48%) and discriminative models performed best for differentiating Alzheimer's disease from controls. The accuracy of algorithms varied with the patient cohort, imaging modalities, and stratifiers used. Few studies performed validation in an independent cohort.

DISCUSSION

The literature has several methodological limitations including lack of sufficient algorithm development descriptions and standard definitions. We make recommendations to improve model validation including addressing key clinical questions, providing sufficient description of AI methods and validating findings in independent datasets. Collaborative approaches between experts in AI and medicine will help achieve the promising potential of AI tools in practice.

HIGHLIGHTS

There has been a rapid expansion in the use of machine learning for diagnosis and prognosis in neurodegenerative disease Most studies (71%) relied on the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset with no other individual dataset used more than five times There has been a recent rise in the use of more complex discriminative models (e.g., neural networks) that performed better than other classifiers for classification of AD vs healthy controls We make recommendations to address methodological considerations, addressing key clinical questions, and validation We also make recommendations for the field more broadly to standardize outcome measures, address gaps in the literature, and monitor sources of bias.

The metabolic spatial covariance pattern of definite idiopathic normal pressure hydrocephalus: an FDG PET study with principal components analysis

Identification of patients with idiopathic normal pressure hydrocephalus (iNPH) in a collective with suspected neurodegenerative disease is essential. This study aimed to determine the metabolic spatial covariance pattern of iNPH on FDG PET using an established technique based on scaled subprofile model principal components analysis (SSM-PCA).We identified 11 patients with definite iNPH. By applying SSM-PCA to the FDG PET data, they were compared to 48 age-matched healthy controls to determine the whole-brain voxel-wise metabolic spatial covariance pattern of definite iNPH (iNPH-related pattern, iNPHRP). The iNPHRP score was compared between groups of patients with definite iNPH, possible iNPH (N = 34), Alzheimer's (AD, N = 38), and Parkinson's disease (PD, N = 35) applying pairwise Mann-Whitney U tests and correction for multiple comparisons.SSM-PCA of FDG PET revealed an iNPHRP that is characterized by relative negative voxel weights at the vicinity of the lateral ventricles and relative positive weights in the paracentral midline region. The iNPHRP scores of patients with definite iNPH were substantially higher than in patients with AD and PD (both p < 0.05) and non-significantly higher than those of patients with possible iNPH. Subject scores of the iNPHRP discriminated definite iNPH from AD and PD with 96% and 100% accuracy and possible iNPH from AD and PD with 83% and 86% accuracy.We defined a novel metabolic spatial covariance pattern of iNPH that might facilitate the differential diagnosis of iNPH versus other neurodegenerative disorders. The knowledge of iNPH-associated alterations in the cerebral glucose metabolism is of high relevance as iNPH constitutes an important differential diagnosis to dementia and movement disorders.

The remodeling of metabolic brain pattern in patients with extracranial diffuse large B-cell lymphoma

BACKGROUND

Owing to the advances in diagnosis and therapy, survival or remission rates for lymphoma have improved prominently. Apart from the lymphoma- and chemotherapy-related somatic symptom burden, increasing attention has been drawn to the health-related quality of life. The application of 18F-fluorodeoxyglucose positron emission tomography-computed tomography (18F-FDG PET/CT) has been routinely recommended for the staging and response assessment of FDG-avid lymphoma. However, up till now, only a few researches have investigated the brain metabolic impairments in patients with pre-treatment lymphoma. The determination of the lymphoma-related metabolic brain pattern would facilitate exploring the tailored therapeutic regimen to alleviate not only the physiological, but also the psychological symptoms. In this retrospective study, we aimed to establish the diffuse large B-cell lymphoma-related pattern (DLBCLRP) of metabolic brain network and investigate the correlations between DLBCLRP and several indexes of the staging and response assessment.

RESULTS

The established DLBCLRP was characterized by the increased metabolic activity in bilateral cerebellum, brainstem, thalamus, striatum, hippocampus, amygdala, parahippocampal gyrus and right middle temporal gyrus and by the decreased metabolic activity in bilateral occipital lobe, parietal lobe, anterior cingulate gyrus, midcingulate cortex and medial frontal gyrus. Significant difference in the baseline expression of DLBCLRP was found among complete metabolic response (CMR), partial metabolic response (PMR) and progressive metabolic disease (PMD) groups (P < 0.01). DLBCLRP expressions were also significantly or tended to be positively correlated with international prognostic index (IPI) (rs = 0.306, P < 0.05), lg(total metabolic tumor volume, TMTV) (r = 0.298, P < 0.05) and lg(total lesion glycolysis, TLG) (r = 0.233, P = 0.064). Though no significant correlation of DLBCLRP expression was found with Ann Arbor staging or tumor SUVmax (P > 0.05), the post-treatment declines of DLBCLRP expression were significantly positively correlated with Ann Arbor staging (rs = 0.284, P < 0.05) and IPI (rs = 0.297, P < 0.05).

CONCLUSIONS

The proposed DLBCLRP would lay the foundation for further investigating the cerebral dysfunction related to DLBCL itself and/or treatments. Besides, the expression of DLBCLRP was associated with the tumor burden of lymphoma, implying a potential biomarker for prognosis.

Longitudinal changes in metabolic network activity in early Alzheimer's disease

INTRODUCTION

The progression of Alzheimer's disease (AD) has been linked to two metabolic networks, the AD-related pattern (ADRP) and the default mode network (DMN).

METHODS

Converting and clinically stable cognitively normal subjects (n = 47) and individuals with mild cognitive impairment (n = 96) underwent 2-[18 F]fluoro-2-deoxy-D-glucose (FDG) positron emission tomography (PET) three or more times over 6 years (nscans  = 705). Expression levels for ADRP and DMN were measured in each subject and time point, and the resulting changes were correlated with cognitive performance. The role of network expression in predicting conversion to dementia was also evaluated.

RESULTS

Longitudinal increases in ADRP expression were observed in converters, while age-related DMN loss was seen in converters and nonconverters. Cognitive decline correlated with increases in ADRP and declines in DMN, but conversion to dementia was predicted only by baseline ADRP levels.

DISCUSSION

The results point to the potential utility of ADRP as an imaging biomarker of AD progression.

Principal component analysis of synaptic density measured with [11C]UCB-J PET in early Alzheimer's disease

BACKGROUND

Synaptic loss is considered an early pathological event and major structural correlate of cognitive impairment in Alzheimer's disease (AD). We used principal component analysis (PCA) to identify regional patterns of covariance in synaptic density using [11C]UCB-J PET and assessed the association between principal components (PC) subject scores with cognitive performance.

METHODS

[11C]UCB-J binding was measured in 45 amyloid + participants with AD and 19 amyloid- cognitively normal participants aged 55-85. A validated neuropsychological battery assessed performance across five cognitive domains. PCA was applied to the pooled sample using distribution volume ratios (DVR) standardized (z-scored) by region from 42 bilateral regions of interest (ROI).

RESULTS

Parallel analysis determined three significant PCs explaining 70.2% of the total variance. PC1 was characterized by positive loadings with similar contributions across the majority of ROIs. PC2 was characterized by positive and negative loadings with strongest contributions from subcortical and parietooccipital cortical regions, respectively, while PC3 was characterized by positive and negative loadings with strongest contributions from rostral and caudal cortical regions, respectively. Within the AD group, PC1 subject scores were positively correlated with performance across all cognitive domains (Pearson r = 0.24-0.40, P = 0.06-0.006), PC2 subject scores were inversely correlated with age (Pearson r = -0.45, P = 0.002) and PC3 subject scores were significantly correlated with CDR-sb (Pearson r = 0.46, P = 0.04). No significant correlations were observed between cognitive performance and PC subject scores in CN participants.

CONCLUSIONS

This data-driven approach defined specific spatial patterns of synaptic density correlated with unique participant characteristics within the AD group. Our findings reinforce synaptic density as a robust biomarker of disease presence and severity in the early stages of AD.

Effect of the identification group size and image resolution on the diagnostic performance of metabolic Alzheimer's disease-related pattern

BACKGROUND

Alzheimer's disease-related pattern (ADRP) is a metabolic brain biomarker of Alzheimer's disease (AD). While ADRP is being introduced into research, the effect of the size of the identification cohort and the effect of the resolution of identification and validation images on ADRP's performance need to be clarified.

METHODS

240 2-[¹⁸F]fluoro-2-deoxy-D-glucose positron emission tomography images [120 AD/120 cognitive normals (CN)] were selected from the Alzheimer's disease neuroimaging initiative database. A total of 200 images (100 AD/100 CN) were used to identify different versions of ADRP using a scaled subprofile model/principal component analysis. For this purpose, five identification groups were randomly selected 25 times. The identification groups differed in the number of images (20 AD/20 CN, 30 AD/30 CN, 40 AD/40 CN, 60 AD/60 CN, and 80 AD/80 CN) and image resolutions (6, 8, 10, 12, 15 and 20 mm). A total of 750 ADRPs were identified and validated through the area under the curve (AUC) values on the remaining 20 AD/20 CN with six different image resolutions.

RESULTS

ADRP's performance for the differentiation between AD patients and CN demonstrated only a marginal average AUC increase, when the number of subjects in the identification group increases (AUC increase for about 0.03 from 20 AD/20 CN to 80 AD/80 CN). However, the average of the lowest five AUC values increased with the increasing number of participants (AUC increase for about 0.07 from 20 AD/20 CN to 30 AD/30 CN and for an additional 0.02 from 30 AD/30 CN to 40 AD/40 CN). The resolution of the identification images affects ADRP's diagnostic performance only marginally in the range from 8 to 15 mm. ADRP's performance stayed optimal even when applied to validation images of resolution differing from the identification images.

CONCLUSIONS

While small (20 AD/20 CN images) identification cohorts may be adequate in a favorable selection of cases, larger cohorts (at least 30 AD/30 CN images) shall be preferred to overcome possible/random biological differences and improve ADRP's diagnostic performance. ADRP's performance stays stable even when applied to the validation images with a resolution different than the resolution of the identification ones.

The identification and cognitive correlation of perfusion patterns measured with arterial spin labeling MRI in Alzheimer's disease

BACKGROUND

Vascular dysfunction, including cerebral hypoperfusion, plays an important role in the pathogenesis and progression of Alzheimer's disease (AD), independent of amyloid and tau pathology. We established an AD-related perfusion pattern (ADRP) measured with arterial spin labeling (ASL) MRI using multivariate spatial covariance analysis.

METHODS

We obtained multimodal MRI including pseudo-continuous ASL and neurocognitive testing in a total of 55 patients with a diagnosis of mild to moderate AD supported by amyloid PET and 46 normal controls (NCs). An ADRP was established from an identification cohort of 32 patients with AD and 32 NCs using a multivariate analysis method based on scaled subprofile model/principal component analysis, and pattern expression in individual subjects was quantified for both the identification cohort and a validation cohort (23 patients with AD and 14 NCs). Subject expression score of the ADRP was then used to assess diagnostic accuracy and cognitive correlations in AD patients and compared with global and regional cerebral blood flow (CBF) in specific areas identified from voxel-based univariate analysis.

RESULTS

The ADRP featured negative loading in the bilateral middle and posterior cingulate and precuneus, inferior parietal lobule, and frontal areas, and positive loading in the right cerebellum and bilateral basal areas. Subject expression score of the ADRP was significantly elevated in AD patients compared with NCs (P < 0.001) and showed good diagnostic accuracy for AD with area under receiver-operator curve of 0.87 [95% CI (0.78-0.96)] in the identification cohort and 0.85 in the validation cohort. Moreover, there were negative correlations between subject expression score and global cognitive function and performance in various cognitive domains in patients with AD. The characteristics of the ADRP topography and subject expression scores were supported by analogous findings obtained with regional CBF.

CONCLUSIONS

We have reported a characteristic perfusion pattern associated with AD using ASL MRI. Subject expression score of this spatial covariance pattern is a promising MRI biomarker for the identification and monitoring of AD.

Inhibiting NLRP3 Inflammasome Activation by CY-09 Helps to Restore Cerebral Glucose Metabolism in 3×Tg-AD Mice

The reduction of the cerebral glucose metabolism is closely related to the activation of the NOD-like receptor protein 3 (NLRP3) inflammasome in Alzheimer’s disease (AD); however, its underlying mechanism remains unclear. In this paper, 18F-flurodeoxyglucose positron emission tomography was used to trace cerebral glucose metabolism in vivo, along with Western blotting and immunofluorescence assays to examine the expression and distribution of associated proteins. Glucose and insulin tolerance tests were carried out to detect insulin resistance, and the Morris water maze was used to test the spatial learning and memory ability of the mice. The results show increased NLRP3 inflammasome activation, elevated insulin resistance, and decreased glucose metabolism in 3×Tg-AD mice. Inhibiting NLRP3 inflammasome activation using CY-09, a specific inhibitor for NLRP3, may restore cerebral glucose metabolism by increasing the expression and distribution of glucose transporters and enzymes and attenuating insulin resistance in AD mice. Moreover, CY-09 helps to improve AD pathology and relieve cognitive impairment in these mice. Although CY-09 has no significant effect on ferroptosis, it can effectively reduce fatty acid synthesis and lipid peroxidation. These findings provide new evidence for NLRP3 inflammasome as a therapeutic target for AD, suggesting that CY-09 may be a potential drug for the treatment of this disease.

Effects of sex and APOE ε4 genotype on brain mitochondrial high-energy phosphates in midlife individuals at risk for Alzheimer's disease: A 31Phosphorus MR spectroscopy study

Age, female sex, and APOE epsilon 4 (APOE4) genotype are the three greatest risk factors for late-onset Alzheimer's disease (AD). The convergence of these risks creates a hypometabolic AD-risk profile unique to women, which may help explain their higher lifetime risk of AD. Less is known about APOE4 effects in men, although APOE4 positive men also experience an increased AD risk. This study uses 31Phosphorus Magnetic Resonance Spectroscopy (31P-MRS) to examine effects of sex and APOE4 status on brain high-energy phosphates [adenosine triphosphate (ATP), phosphocreatine (PCr), inorganic phosphate (Pi)] and membrane phospholipids [phosphomonoesters (PME), phosphodiesters (PDE)] in 209 cognitively normal individuals at risk for AD, ages 40-65, 80% female, 46% APOE4 carriers (APOE4+). Women exhibited lower PCr/ATP and PCr/Pi levels than men in AD-vulnerable regions, including frontal, posterior cingulate, lateral and medial temporal cortex (multi-variable adjusted p≤0.037). The APOE4+ group exhibited lower PCr/ATP and PCr/Pi in frontal regions as compared to non-carriers (APOE4-) (multi-variable adjusted p≤0.005). Sex by APOE4 status interactions were observed in frontal regions (multi-variable adjusted p≤0.046), where both female groups and APOE4+ men exhibited lower PCr/ATP and PCr/Pi than APOE4- men. Among men, APOE4 homozygotes exhibited lower frontal PCr/ATP than heterozygotes and non-carriers. There were no significant effects of sex or APOE4 status on Pi/ATP and PME/PDE measures. Among midlife individuals at risk for AD, women exhibit lower PCr/ATP (e.g. higher ATP utilization) and lower PCr/Pi (e.g. higher energy demand) than age-controlled men, independent of APOE4 status. However, a double dose of APOE4 allele shifted men's brains to a similar metabolic range as women's brains. Examination of brain metabolic heterogeneity can support identification of AD-specific pathways within at-risk subgroups, further advancing both preventive and precision medicine for AD.

Functional brain networks in the evaluation of patients with neurodegenerative disorders

Network analytical tools are increasingly being applied to brain imaging maps of resting metabolic activity (PET) or blood oxygenation-dependent signals (functional MRI) to characterize the abnormal neural circuitry that underlies brain diseases. This approach is particularly valuable for the study of neurodegenerative disorders, which are characterized by stereotyped spread of pathology along discrete neural pathways. Identification and validation of disease-specific brain networks facilitate the quantitative assessment of pathway changes over time and during the course of treatment. Network abnormalities can often be identified before symptom onset and can be used to track disease progression even in the preclinical period. Likewise, network activity can be modulated by treatment and might therefore be used as a marker of efficacy in clinical trials. Finally, early differential diagnosis can be achieved by simultaneously measuring the activity levels of multiple disease networks in an individual patient's scans. Although these techniques were originally developed for PET, over the past several years analogous methods have been introduced for functional MRI, a more accessible non-invasive imaging modality. This advance is expected to broaden the application of network tools to large and diverse patient populations.

Spatially informed Bayesian neural network for neurodegenerative diseases classification

Magnetic resonance imaging (MRI) plays an increasingly important role in the diagnosis and prognosis of neurodegenerative diseases. One field of extensive clinical use of MRI is the accurate and automated classification of degenerative disorders. Most of current classification studies either do not mirror medical practice where patients may exhibit early stages of the disease, comorbidities, or atypical variants, or they are not able to produce probabilistic predictions nor account for uncertainty. Also, the spatial heterogeneity of the brain alterations caused by neurodegenerative processes is not usually considered, despite the spatial configuration of the neuronal loss is a characteristic hallmark for each disorder. In this article, we propose a classification technique that incorporates uncertainty and spatial information for distinguishing between healthy subjects and patients from four distinct neurodegenerative diseases: Alzheimer's disease, mild cognitive impairment, Parkinson's disease, and Multiple Sclerosis. We introduce a spatially informed Bayesian neural network (SBNN) that combines a three-dimensional neural network to extract neurodegeneration features from MRI, Bayesian inference to account for uncertainty in diagnosis, and a spatially informed MRI image using hidden Markov random fields to encode cerebral spatial information. The SBNN model demonstrates that classification accuracy increases up to 25% by including a spatially informed MRI scan. Furthermore, the SBNN provides a robust probabilistic diagnosis that resembles clinical decision-making and can account for the heterogeneous medical presentations of neurodegenerative disorders.

Automated differential diagnosis of dementia syndromes using FDG PET and machine learning

BACKGROUND

Metabolic brain imaging with 2-[¹⁸F]fluoro-2-deoxy-D-glucose positron emission tomography (FDG PET) is a supportive diagnostic and differential diagnostic tool for neurodegenerative dementias. In the clinic, scans are usually visually interpreted. However, computer-aided approaches can improve diagnostic accuracy. We aimed to build two machine learning classifiers, based on two sets of FDG PET-derived features, for differential diagnosis of common dementia syndromes.

METHODS

We analyzed FDG PET scans from three dementia cohorts [63 dementia due to Alzheimer's disease (AD), 79 dementia with Lewy bodies (DLB) and 23 frontotemporal dementia (FTD)], and 41 normal controls (NCs). Patients' clinical diagnosis at follow-up (25 ± 20 months after scanning) or cerebrospinal fluid biomarkers for Alzheimer's disease was considered a gold standard. FDG PET scans were first visually evaluated. Scans were pre-processed, and two sets of features extracted: (1) the expressions of previously identified metabolic brain patterns, and (2) the mean uptake value in 95 regions of interest (ROIs). Two multi-class support vector machine (SVM) classifiers were tested and their diagnostic performance assessed and compared to visual reading. Class-specific regional feature importance was assessed with Shapley Additive Explanations.

RESULTS

Pattern- and ROI-based classifier achieved higher overall accuracy than expert readers (78% and 80% respectively, vs. 71%). Both SVM classifiers performed similarly to one another and to expert readers in AD (F1 = 0.74, 0.78, and 0.78) and DLB (F1 = 0.81, 0.81, and 0.78). SVM classifiers outperformed expert readers in FTD (F1 = 0.87, 0.83, and 0.63), but not in NC (F1 = 0.71, 0.75, and 0.92). Visualization of the SVM model showed bilateral temporal cortices and cerebellum to be the most important features for AD; occipital cortices, hippocampi and parahippocampi, amygdala, and middle temporal lobes for DLB; bilateral frontal cortices, middle and anterior cingulum for FTD; and bilateral angular gyri, pons, and vermis for NC.

CONCLUSION

Multi-class SVM classifiers based on the expression of characteristic metabolic brain patterns or ROI glucose uptake, performed better than experts in the differential diagnosis of common dementias using FDG PET scans. Experts performed better in the recognition of normal scans and a combined approach may yield optimal results in the clinical setting.

Cognitive and Brain Metabolism Profiles of Mild Cognitive Impairment in Prodromal Alpha-Synucleinopathy

Background: Mild cognitive impairment (MCI) is a heterogeneous condition. Idiopathic REM sleep behavior disorder (iRBD) can be associated with MCI (MCI-RBD). Objective: To investigate neuropsychological and brain metabolism features of patients with MCI-RBD by comparison with matched MCI-AD patients. To explore their predictive value toward conversion to a full-blown neurodegenerative disease. Methods: Seventeen MCI-RBD patients (73.6±6.5 years) were enrolled. Thirty-four patients with MCI-AD were matched for age (74.8±4.4 years), Mini-Mental State Exam score and education with a case-control criterion. All patients underwent a neuropsychological assessment and brain 18F-FDG-PET. Images were compared between groups to identify hypometabolic volumes of interest (MCI-RBD-VOI and MCI-AD-VOI). The dependency of whole-brain scaled metabolism levels in MCI-RBD-VOI and MCI-AD-VOI on neuropsychological test scores was explored with linear regression analyses in both groups, adjusting for age and education. Survival analysis was performed to investigate VOIs phenoconversion prediction power. Results: MCI-RBD group scored lower in executive functions and higher in verbal memory compared to MCI-AD group. Also, compared with MCI-AD, MCI-RBD group showed relative hypometabolism in a posterior brain area including cuneus, precuneus, and occipital regions while the inverse comparison revealed relative hypometabolism in the hippocampus/parahippocampal areas in MCI-AD group. MCI-RBD-VOI metabolism directly correlated with executive functions in MCI-RBD (p = 0.04). MCI-AD-VOI metabolism directly correlated with verbal memory in MCI-AD (p = 0.001). MCI-RBD-VOI metabolism predicted (p = 0.03) phenoconversion to an alpha-synucleinopathy. MCI-AD-VOI metabolism showed a trend (p = 0.07) in predicting phenoconversion to dementia. Conclusion: MCI-RBD and MCI-AD showed distinct neuropsychological and brain metabolism profiles, that may be helpful for both diagnosis and prognosis purposes.

Consistent Abnormalities in Metabolic Patterns of Lewy Body Dementias

BACKGROUND

Whether dementia with Lewy bodies (DLB) and Parkinson's disease (PD) dementia (PDD) represent the same disease, distinct entities, or conditions within the same spectrum remains controversial.

OBJECTIVE

The objective of this study was to provide new insight into this debate by separately identifying disease-specific metabolic patterns and comparing them with each other and with previously established PD-related pattern (PDRP).

METHODS

Patients with DLB (n = 67), patients with PDD (n = 50), and healthy control subjects (HCs; n = 15) with brain ¹⁸ F-fluorodeoxyglucose positron emission tomography were enrolled as cohorts A and B for pattern identification and validation, respectively. Patients with PD (n = 30) were included for discrimination. Twenty-one participants had two scans. The principal component analysis was applied for pattern identification (DLB-related pattern [DLBRP], PDD-related pattern [PDDRP]). Similarities and differences among three patterns were assessed by pattern topography, pattern expression, clinical correlations cross-sectionally, and pattern expression changes longitudinally.

RESULTS

DLBRP and PDDRP shared highly similar topographies, with relative hypometabolism mainly in the middle temporal gyrus, middle occipital gyrus, lingual gyrus, precuneus, cuneus, angular gyrus, superior and inferior parietal gyrus, middle and inferior frontal gyrus, cingulate, and caudate, and relative hypermetabolism in the cerebellum, putamen, thalamus, precentral/postcentral gyrus, and paracentral lobule, which were more extensive than the PDRP. Patients with DLB and PDD could not be distinguished successfully by any pattern, but patients with PD were easily recognized, especially by DLBRP and PDDRP. The pattern expression of DLBRP and PDDRP showed similar efficacy in cross-sectional disease severity assessment and longitudinal progression monitoring.

CONCLUSIONS

The consistent abnormalities in metabolic patterns of DLB and PDD might underline the potential continuum across the clinical spectrum from PD to DLB. © 2022 International Parkinson and Movement Disorder Society.

Using radiomics-based modelling to predict individual progression from mild cognitive impairment to Alzheimer's disease

BACKGROUND

Predicting the risk of disease progression from mild cognitive impairment (MCI) to Alzheimer's disease (AD) has important clinical significance. This study aimed to provide a personalized MCI-to-AD conversion prediction via radiomics-based predictive modelling (RPM) with multicenter 18F-fluorodeoxyglucose positron emission tomography (FDG PET) data.

METHOD

FDG PET and neuropsychological data of 884 subjects were collected from Huashan Hospital, Xuanwu Hospital, and from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database. First, 34,400 radiomic features were extracted from the 80 regions of interest (ROIs) for all PET images. These features were then concatenated for feature selection, and an RPM model was constructed and validated on the ADNI dataset. In addition, we used clinical data and the routine semiquantification index (standard uptake value ratio, SUVR) to establish clinical and SUVR Cox models for further comparison. FDG images from local hospitals were used to explore RPM performance in a separate cohort of individuals with healthy controls and different cognitive levels (a complete AD continuum). Finally, correlation analysis was conducted between the radiomic biomarkers and neuropsychological assessments.

RESULTS

The experimental results showed that the predictive performance of the RPM Cox model was better than that of other Cox models. In the validation dataset, Harrell's consistency coefficient of the RPM model was 0.703 ± 0.002, while those of the clinical and SUVR models were 0.632 ± 0.006 and 0.683 ± 0.009, respectively. Moreover, most crucial imaging biomarkers were significantly different at different cognitive stages and significantly correlated with cognitive disease severity.

CONCLUSION

The preliminary results demonstrated that the developed RPM approach has the potential to monitor progression in high-risk populations with AD.

BMNet: A New Region-Based Metric Learning Method for Early Alzheimer's Disease Identification With FDG-PET Images

18F-fluorodeoxyglucose (FDG)-positron emission tomography (PET) reveals altered brain metabolism in individuals with mild cognitive impairment (MCI) and Alzheimer's disease (AD). Some biomarkers derived from FDG-PET by computer-aided-diagnosis (CAD) technologies have been proved that they can accurately diagnosis normal control (NC), MCI, and AD. However, existing FDG-PET-based researches are still insufficient for the identification of early MCI (EMCI) and late MCI (LMCI). Compared with methods based other modalities, current methods with FDG-PET are also inadequate in using the inter-region-based features for the diagnosis of early AD. Moreover, considering the variability in different individuals, some hard samples which are very similar with both two classes limit the classification performance. To tackle these problems, in this paper, we propose a novel bilinear pooling and metric learning network (BMNet), which can extract the inter-region representation features and distinguish hard samples by constructing the embedding space. To validate the proposed method, we collect 898 FDG-PET images from Alzheimer's disease neuroimaging initiative (ADNI) including 263 normal control (NC) patients, 290 EMCI patients, 147 LMCI patients, and 198 AD patients. Following the common preprocessing steps, 90 features are extracted from each FDG-PET image according to the automatic anatomical landmark (AAL) template and then sent into the proposed network. Extensive fivefold cross-validation experiments are performed for multiple two-class classifications. Experiments show that most metrics are improved after adding the bilinear pooling module and metric losses to the Baseline model respectively. Specifically, in the classification task between EMCI and LMCI, the specificity improves 6.38% after adding the triple metric loss, and the negative predictive value (NPV) improves 3.45% after using the bilinear pooling module. In addition, the accuracy of classification between EMCI and LMCI achieves 79.64% using imbalanced FDG-PET images, which illustrates that the proposed method yields a state-of-the-art result of the classification accuracy between EMCI and LMCI based on PET images.

Metabolic imaging and plasticity

Positron emission tomography greatly advanced our understanding on the underlying neural mechanisms of movement disorders. PET with flurodeoxyglucose (FDG) is especially useful as it depicts regional metabolic activity level that can predict patients' symptoms. Multivariate pattern analysis has been used to determine and quantify the co-varying brain networks associated with specific clinical traits of neurodegenerative disease. The result is a biomarker, useful for diagnosis, treatments, and follow up studies. Parkinsonian traits and parkinsonisms are associated with specific spatial pattern of metabolic abnormality useful for differential diagnosis. This approach has also been used for monitoring disease progression and novel treatment responses mostly in Parkinson's disease. In this book chapter, we, illustrate and discuss the significance of the brain networks associated with disease and their modification with neuroplastic changes.

Exploring brain glucose metabolic patterns in cognitively normal adults at risk of Alzheimer's disease: A cross-validation study with Chinese and ADNI cohorts

OBJECTIVE

Disease-related metabolic brain patterns have been verified for a variety of neurodegenerative diseases including Alzheimer's disease (AD). This study aimed to explore and validate the pattern derived from cognitively normal controls (NCs) in the Alzheimer's continuum.

METHODS

This study was based on two cohorts; one from the Alzheimer's Disease Neuroimaging Initiative (ADNI) and the other from the Sino Longitudinal Study on Cognitive Decline (SILCODE). Each subject underwent [18F]fluoro-2-deoxyglucose positron emission tomography (PET) and [18F]florbetapir-PET imaging. Participants were binary-grouped based on β-amyloid (Aβ) status, and the positivity was defined as Aβ+. Voxel-based scaled subprofile model/principal component analysis (SSM/PCA) was used to generate the "at-risk AD-related metabolic pattern (ARADRP)" for NCs. The pattern expression score was obtained and compared between the groups, and receiver operating characteristic curves were drawn. Notably, we conducted cross-validation to verify the robustness and correlation analyses to explore the relationships between the score and AD-related pathological biomarkers.

RESULTS

Forty-eight Aβ+ NCs and 48 Aβ- NCs were included in the ADNI cohort, and 25 Aβ+ NCs and 30 Aβ- NCs were included in the SILCODE cohort. The ARADRPs were identified from the combined cohorts and the two separate cohorts, characterized by relatively lower regional loadings in the posterior parts of the precuneus, posterior cingulate, and regions of the temporal gyrus, as well as relatively higher values in the superior/middle frontal gyrus and other areas. Patterns identified from the two separate cohorts showed some regional differences, including the temporal gyrus, basal ganglia regions, anterior parts of the precuneus, and middle cingulate. Cross-validation suggested that the pattern expression score was significantly higher in the Aβ+ group of both cohorts (p < 0.01), and contributed to the diagnosis of Aβ+ NCs (with area under the curve values of 0.696-0.815). The correlation analysis revealed that the score was related to tau pathology measured in cerebrospinal fluid (p-tau: p < 0.02; t-tau: p < 0.03), but not Aβ pathology assessed with [18F]florbetapir-PET (p > 0.23).

CONCLUSIONS

ARADRP exists for NCs, and the acquired pattern expression score shows a certain ability to discriminate Aβ+ NCs from Aβ- NCs. The SSM/PCA method is expected to be helpful in the ultra-early diagnosis of AD in clinical practice.

Sex differences in Alzheimer's disease: metabolic reprogramming and therapeutic intervention

Studies on the sporadic form of Alzheimer's disease (AD) have revealed three classes of risk factor: age, genetics, and sex. These risk factors point to a metabolic dysregulation as the origin of AD. Adaptive alterations in cerebral metabolism are the rationale for the Metabolic Reprogramming (MR) Theory of the origin of AD. The theory contends that the progression toward AD involves three adaptive events: a hypermetabolic phase, a prolonged prodromal phase, and a metabolic collapse. This article exploits the MR Theory to elucidate the effect of hormonal changes on the origin and progression of AD in women. The theory invokes bioenergetic signatures of the menopausal transition to propose sex-specific diagnostic program and therapeutic strategies.

Application of Data Mining Algorithms for Dementia in People with HIV/AIDS

Dementia interferes with the individual's motor, behavioural, and intellectual functions, causing him to be unable to perform instrumental activities of daily living. This study is aimed at identifying the best performing algorithm and the most relevant characteristics to categorise individuals with HIV/AIDS at high risk of dementia from the application of data mining. Principal component analysis (PCA) algorithm was used and tested comparatively between the following machine learning algorithms: logistic regression, decision tree, neural network, KNN, and random forest. The database used for this study was built from the data collection of 270 individuals infected with HIV/AIDS and followed up at the outpatient clinic of a reference hospital for infectious and parasitic diseases in the State of Ceará, Brazil, from January to April 2019. Also, the performance of the algorithms was analysed for the 104 characteristics available in the database; then, with the reduction of dimensionality, there was an improvement in the quality of the machine learning algorithms and identified that during the tests, even losing about 30% of the variation. Besides, when considering only 23 characteristics, the precision of the algorithms was 86% in random forest, 56% logistic regression, 68% decision tree, 60% KNN, and 59% neural network. The random forest algorithm proved to be more effective than the others, obtaining 84% precision and 86% accuracy.

A disease-specific metabolic imaging marker for diagnosis and progression evaluation of semantic variant primary progressive aphasia

BACKGROUND AND PURPOSE

The diagnosis and monitoring of semantic variant primary progressive aphasia (sv-PPA) are clinically challenging. We aimed to establish a distinctive metabolic pattern in sv-PPA for diagnosis and severity evaluation.

METHODS

Fifteen sv-PPA patients and 15 controls were enrolled to identify sv-PPA-related pattern (sv-PPARP) by principal component analysis of ¹⁸ F-fluorodeoxyglucose positron emission tomography. Eighteen Alzheimer disease dementia (AD) and 14 behavioral variant frontotemporal dementia (bv-FTD) patients were enrolled to test the discriminatory power. Correspondingly, regional metabolic activities extracted from the voxelwise analysis were evaluated for the discriminatory power.

RESULTS

The sv-PPARP was characterized as decreased metabolic activity mainly in the bilateral temporal lobe (left predominance), middle orbitofrontal gyrus, left hippocampus/parahippocampus gyrus, fusiform gyrus, insula, inferior orbitofrontal gyrus, and striatum, with increased activity in the bilateral lingual gyrus, cuneus, calcarine gyrus, and right precentral and postcentral gyrus. The pattern expression had significant discriminatory power (area under the curve [AUC] = 0.98, sensitivity = 100%, specificity = 94.4%) in distinguishing sv-PPA from AD, and the asymmetry index offered complementary discriminatory power (AUC = 0.91, sensitivity = 86.7%, specificity = 92.9%) in distinguishing sv-PPA from bv-FTD. In sv-PPA patients, the pattern expression correlated with Boston Naming Test scores at baseline and showed significant increase in the subset of patients with follow-up. The voxelwise analysis showed similar topography, and the regional metabolic activities had equivalent or better discriminatory power and clinical correlations with Boston Naming Test scores. The ability to reflect disease progression in longitudinal follow-up seemed to be inferior to the pattern expression.

CONCLUSIONS

The sv-PPARP might serve as an objective biomarker for diagnosis and progression evaluation.

Cognitive impairment and altered cerebral glucose metabolism in the subacute stage of COVID-19

During the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic, neurological symptoms increasingly moved into the focus of interest. In this prospective cohort study, we assessed neurological and cognitive symptoms in hospitalized coronavirus disease-19 (COVID-19) patients and aimed to determine their neuronal correlates. Patients with reverse transcription-PCR-confirmed COVID-19 infection who required inpatient treatment primarily because of non-neurological complications were screened between 20 April 2020 and 12 May 2020. Patients (age > 18 years) were included in our cohort when presenting with at least one new neurological symptom (defined as impaired gustation and/or olfaction, performance < 26 points on a Montreal Cognitive Assessment and/or pathological findings on clinical neurological examination). Patients with ≥2 new symptoms were eligible for further diagnostics using comprehensive neuropsychological tests, cerebral MRI and ¹⁸fluorodeoxyglucose (FDG) PET as soon as infectivity was no longer present. Exclusion criteria were: premorbid diagnosis of cognitive impairment, neurodegenerative diseases or intensive care unit treatment. Of 41 COVID-19 inpatients screened, 29 patients (65.2 ± 14.4 years; 38% female) in the subacute stage of disease were included in the register. Most frequently, gustation and olfaction were disturbed in 29/29 and 25/29 patients, respectively. Montreal Cognitive Assessment performance was impaired in 18/26 patients (mean score 21.8/30) with emphasis on frontoparietal cognitive functions. This was confirmed by detailed neuropsychological testing in 15 patients. ¹⁸FDG PET revealed pathological results in 10/15 patients with predominant frontoparietal hypometabolism. This pattern was confirmed by comparison with a control sample using voxel-wise principal components analysis, which showed a high correlation (R² = 0.62) with the Montreal Cognitive Assessment performance. Post-mortem examination of one patient revealed white matter microglia activation but no signs of neuroinflammation. Neocortical dysfunction accompanied by cognitive decline was detected in a relevant fraction of patients with subacute COVID-19 initially requiring inpatient treatment. This is of major rehabilitative and socioeconomic relevance.

Multitask fMRI and machine learning approach improve prediction of differential brain activity pattern in patients with insomnia disorder

We investigated the differential spatial covariance pattern of blood oxygen level-dependent (BOLD) responses to single-task and multitask functional magnetic resonance imaging (fMRI) between patients with psychophysiological insomnia (PI) and healthy controls (HCs), and evaluated features generated by principal component analysis (PCA) for discrimination of PI from HC, compared to features generated from BOLD responses to single-task fMRI using machine learning methods. In 19 patients with PI and 21 HCs, the mean beta value for each region of interest (ROIbval) was calculated with three contrast images (i.e., sleep-related picture, sleep-related sound, and Stroop stimuli). We performed discrimination analysis and compared with features generated from BOLD responses to single-task fMRI. We applied support vector machine analysis with a least absolute shrinkage and selection operator to evaluate five performance metrics: accuracy, recall, precision, specificity, and F2. Principal component features showed the best classification performance in all aspects of metrics compared to BOLD response to single-task fMRI. Bilateral inferior frontal gyrus (orbital), right calcarine cortex, right lingual gyrus, left inferior occipital gyrus, and left inferior temporal gyrus were identified as the most salient areas by feature selection. Our approach showed better performance in discriminating patients with PI from HCs, compared to single-task fMRI.

Neuroimaging evaluation of deep brain stimulation in the treatment of representative neurodegenerative and neuropsychiatric disorders

Brain stimulation technology has become a viable modality of reversible interventions in the effective treatment of many neurological and psychiatric disorders. It is aimed to restore brain dysfunction by the targeted delivery of specific electronic signal within or outside the brain to modulate neural activity on local and circuit levels. Development of therapeutic approaches with brain stimulation goes in tandem with the use of neuroimaging methodology in every step of the way. Indeed, multimodality neuroimaging tools have played important roles in target identification, neurosurgical planning, placement of stimulators and post-operative confirmation. They have also been indispensable in pre-treatment screen to identify potential responders and in post-treatment to assess the modulation of brain circuitry in relation to clinical outcome measures. Studies in patients to date have elucidated novel neurobiological mechanisms underlying the neuropathogenesis, action of stimulations, brain responses and therapeutic efficacy. In this article, we review some applications of deep brain stimulation for the treatment of several diseases in the field of neurology and psychiatry. We highlight how the synergistic combination of brain stimulation and neuroimaging technology is posed to accelerate the development of symptomatic therapies and bring revolutionary advances in the domain of bioelectronic medicine.

Neuroimage Biomarker Identification of the Conversion of Mild Cognitive Impairment to Alzheimer's Disease

An efficient method to identify whether mild cognitive impairment (MCI) has progressed to Alzheimer's disease (AD) will be beneficial to patient care. Previous studies have shown that magnetic resonance imaging (MRI) has enabled the assessment of AD progression based on imaging findings. The present work aimed to establish an algorithm based on three features, namely, volume, surface area, and surface curvature within the hippocampal subfields, to model variations, including atrophy and structural changes to the cortical surface. In this study, a new biomarker, the ratio of principal curvatures (RPC), was proposed to characterize the folding patterns of the cortical gyrus and sulcus. Along with volumes and surface areas, these morphological features associated with the hippocampal subfields were assessed in terms of their sensitivity to the changes in cognitive capacity by two different feature selection methods. Either the extracted features were statistically significantly different, or the features were selected through a random forest model. The identified subfields and their structural indices that are sensitive to the changes characteristic of the progression from MCI to AD were further assessed with a multilayer perceptron classifier to help facilitate the diagnosis. The accuracy of the classification based on the proposed method to distinguish whether a MCI patient enters the AD stage amounted to 79.95%, solely using the information from the features selected by a logical feature selection method.

Validation of the Alzheimer Disease Dementia Conversion-Related Pattern as an ATN Biomarker of Neurodegeneration

OBJECTIVE

To determine whether the Alzheimer disease (AD) dementia conversion-related pattern (ADCRP) on [¹⁸F]FDG PET can serve as a valid predictor for the development of AD dementia, the individual expression of the ADCRP (subject score) and its prognostic value were examined in patients with mild cognitive impairment (MCI) and biologically defined AD.

METHODS

A total of 269 patients with available [¹⁸F]FDG PET, [¹⁸F]AV-45 PET, phosphorylated and total tau in CSF, and neurofilament light chain in plasma were included. Following the AT(N) classification scheme, where AD is defined biologically by in vivo biomarkers of β-amyloid (Aβ) deposition ("A") and pathologic tau ("T"), patients were categorized to the A-T-, A+T-, A+T+ (AD), and A-T+ groups.

RESULTS

The mean subject score of the ADCRP was significantly higher in the A+T+ group compared to each of the other group (all p < 0.05) but was similar among the latter (all p > 0.1). Within the A+T+ group, the subject score of ADCRP was a significant predictor of conversion to dementia (hazard ratio, 2.02 per z score increase; p < 0.001), with higher predictive value than of alternative biomarkers of neurodegeneration (total tau and neurofilament light chain). Stratification of A+T+ patients by the subject score of ADCRP yielded well-separated groups of high, medium, and low conversion risks.

CONCLUSIONS

The ADCRP is a valuable biomarker of neurodegeneration in patients with MCI and biologically defined AD. It shows great potential for stratifying the risk and estimating the time to conversion to dementia in patients with MCI and underlying AD (A+T+).

CLASSIFICATION OF EVIDENCE

This study provides Class I evidence that [¹⁸F]FDG PET predicts the development of AD dementia in individuals with MCI and underlying AD as defined by the AT(N) framework.

Principal-Component Analysis-Based Measures of PET Data Closely Reflect Neuropathologic Staging Schemes

Voxel-based principal-component analysis allows for an identification of patterns of glucose metabolism and amyloid deposition related to the conversion from mild cognitive impairment (MCI) to Alzheimer disease (AD). The present study aimed to validate these AD conversion-related patterns (ADCRPs) against neuropathologic findings. Methods: We included patients from the Alzheimer's Disease Neuroimaging Initiative who underwent autopsy and for whom ¹⁸F-FDG PET (30 AD, 6 MCI, 2 cognitively normal) and amyloid-β (Aβ) PET (17 AD, 3 MCI, 2 cognitively normal) were available. Pattern expression scores (PESs) of the ¹⁸F-FDG- and Aβ-ADCRP were compared with Braak tangle stage and Thal amyloid phase, respectively. Mean ¹⁸F-FDG uptake and mean ¹⁸F-AV-45 SUV ratio (SUVr) in regions of hypometabolism and elevated amyloid load typical of AD, respectively, were used as volume-of-interest-based PET measures. The diagnostic performance for identifying none-to-low vs. intermediate-to-high AD neuropathologic change (ADNC) was assessed for all biomarkers. Results: We observed significant associations between PES of ¹⁸F-FDG-ADCRP and Braak stage (ρ > 0.48, P < 0.005) and between PES of Aβ-ADCRP and Thal phase (ρ > 0.66, P < 0.001). PES of ¹⁸F-FDG-ADCRP, PES of Aβ-ADCRP, and their combination identified intermediate-to-high ADNC with an area under the receiver-operating-characteristic curve (AUC) of 0.80, 0.95, and 0.98 (n = 22), respectively. Mean ¹⁸F-FDG uptake and mean ¹⁸F-AV-45 SUVr in AD-typical regions were also significantly associated with Braak stage (|ρ| > 0.45, P < 0.01) and Thal phase (ρ > 0.55, P < 0.01), respectively. Volume-of-interest-based PET measures discriminated between ADNC stages with an AUC of 0.79, 0.88, and 0.90 for mean ¹⁸F-FDG uptake, mean ¹⁸F-AV-45 SUVr, and their combination (n = 22), respectively. Contemplating all subjects with available ¹⁸F-FDG PET and neuropathology information (n = 38), PES of ¹⁸F-FDG-ADCRP was a significant predictor of intermediate-to-high ADNC (AUC = 0.72), whereas mean ¹⁸F-FDG uptake was not (AUC = 0.66), although the difference between methods was not significant. Conclusion: PES of ¹⁸F-FDG-ADCRP, a measure of neurodegeneration, shows close correspondence with the extent of tau pathology, as assessed by Braak tangle stage. PES of Aβ-ADCRP is a valid biomarker of underlying amyloid pathology, as demonstrated by its strong correlation with Thal phase. The combination of ADCRPs performed better than ¹⁸F-FDG-ADCRP alone, although there was only negligible improvement compared with Aβ-ADCRP.

Brain amyloid β, cerebral small vessel disease, and cognition: A memory clinic study

OBJECTIVE

To evaluate the association between brain amyloid β (Aβ) and cerebral small vessel disease (CSVD) markers, as well as their joint effect on cognition, in a memory clinic study.

METHODS

A total of 186 individuals visiting a memory clinic, diagnosed with no cognitive impairment, cognitive impairment no dementia (CIND), Alzheimer dementia (AD), or vascular dementia were included. Brain Aβ was measured by [¹¹C] Pittsburgh compound B-PET global standardized uptake value ratio (SUVR). CSVD markers including white matter hyperintensities (WMH), lacunes, and cerebral microbleeds (CMBs) were graded on MRI. Cognition was assessed by neuropsychological testing.

RESULTS

An increase in global SUVR is associated with a decrease in Mini-Mental State Examination (MMSE) in CIND and AD, as well as a decrease in global cognition Z score in AD, independent of age, education, hippocampal volume, and markers of CSVD. A significant interaction between global SUVR and WMH was found in relation to MMSE in CIND (P for interaction: 0.009), with an increase of the effect size of Aβ (β = -6.57 [-9.62 to -3.54], p < 0.001) compared to the model without the interaction term (β = -2.91 [-4.54 to -1.29], p = 0.001).

CONCLUSION

Higher global SUVR was associated with worse cognition in CIND and AD, but was augmented by an interaction between global SUVR and WMH only in CIND. This suggests that Aβ and CSVD are independent processes with a possible synergistic effect between Aβ and WMH in individuals with CIND. There was no interaction effect between Aβ and lacunes or CMBs. Therefore, in preclinical phases of AD, WMH should be targeted as a potentially modifiable factor to prevent worsening of cognitive dysfunction.

The Use of Random Forests to Identify Brain Regions on Amyloid and FDG PET Associated With MoCA Score

PURPOSE

The aim of this study was to evaluate random forests (RFs) to identify ROIs on F-florbetapir and F-FDG PET associated with Montreal Cognitive Assessment (MoCA) score.

MATERIALS AND METHODS

Fifty-seven subjects with significant white matter disease presenting with either transient ischemic attack/lacunar stroke or mild cognitive impairment from early Alzheimer disease, enrolled in a multicenter prospective observational trial, had MoCA and F-florbetapir PET; 55 had F-FDG PET. Scans were processed using the MINC toolkit to generate SUV ratios, normalized to cerebellar gray matter (F-florbetapir PET), or pons (F-FDG PET). SUV ratio data and MoCA score were used for supervised training of RFs programmed in MATLAB.

RESULTS

F-Florbetapir PETs were randomly divided into 40 training and 17 testing scans; 100 RFs of 1000 trees, constructed from a random subset of 16 training scans and 20 ROIs, identified ROIs associated with MoCA score: right posterior cingulate gyrus, right anterior cingulate gyrus, left precuneus, left posterior cingulate gyrus, and right precuneus. Amyloid increased with decreasing MoCA score. F-FDG PETs were randomly divided into 40 training and 15 testing scans; 100 RFs of 1000 trees, each tree constructed from a random subset of 16 training scans and 20 ROIs, identified ROIs associated with MoCA score: left fusiform gyrus, left precuneus, left posterior cingulate gyrus, right precuneus, and left middle orbitofrontal gyrus. F-FDG decreased with decreasing MoCA score.

CONCLUSIONS

Random forests help pinpoint clinically relevant ROIs associated with MoCA score; amyloid increased and F-FDG decreased with decreasing MoCA score, most significantly in the posterior cingulate gyrus.

A Novel Metabolic Connectome Method to Predict Progression to Mild Cognitive Impairment

OBJECTIVE

Glucose-based positron emission tomography (PET) imaging has been widely used to predict the progression of mild cognitive impairment (MCI) into Alzheimer's disease (AD) clinically. However, existing discriminant methods are unsubtle to reveal pathophysiological changes. Therefore, we present a novel metabolic connectome-based predictive modeling to predict progression from MCI to AD accurately.

METHODS

In this study, we acquired fluorodeoxyglucose PET images and clinical assessments from 420 MCI patients with 36 months follow-up. Individual metabolic network based on connectome analysis was constructed, and the metabolic connectivity in this network was extracted as predictive features. Three different classification strategies were implemented to interrogate the predictive performance. To verify the effectivity of selected features, specific brain regions associated with MCI conversion were identified based on these features and compared with prior knowledge.

RESULTS

As a result, 4005 connectome features were obtained, and 153 in which were selected as efficient features. Our proposed feature extraction method had achieved 85.2% accuracy for MCI conversion prediction (sensitivity: 88.1%; specificity: 81.2%; and AUC: 0.933). The discriminative brain regions associated with MCI conversion were mainly located in the precentral gyrus, precuneus, lingual, and inferior frontal gyrus.

CONCLUSION

Overall, the results suggest that our proposed individual metabolic connectome method has great potential to predict whether MCI patients will progress to AD. The metabolic connectome may help to identify brain metabolic dysfunction and build a clinically applicable biomarker to predict the MCI progression.

Associations of [18F]-APN-1607 Tau PET Binding in the Brain of Alzheimer's Disease Patients With Cognition and Glucose Metabolism

Molecular imaging of tauopathies is complicated by the differing specificities and off-target binding properties of available radioligands for positron emission tomography (PET). [18F]-APN-1607 ([18F]-PM-PBB3) is a newly developed PET tracer with promising properties for tau imaging. We aimed to characterize the cerebral binding of [18F]-APN-1607 in Alzheimer's disease (AD) patients compared to normal control (NC) subjects. Therefore, we obtained static late frame PET recordings with [18F]-APN-1607 and [18F]-FDG in patients with a clinical diagnosis of AD group, along with an age-matched NC group ([18F]-APN-1607 only). Using statistical parametric mapping (SPM) and volume of interest (VOI) analyses of the reference region normalized standardized uptake value ratio maps, we then tested for group differences and relationships between both PET biomarkers, as well as their associations with clinical general cognition. In the AD group, [18F]-APN-1607 binding was elevated in widespread cortical regions (P < 0.001 for VOI analysis, familywise error-corrected P < 0.01 for SPM analysis). The regional uptake in AD patients correlated negatively with Mini-Mental State Examination score (frontal lobe: R = -0.632, P = 0.004; temporal lobe: R = -0.593, P = 0.008; parietal lobe: R = -0.552, P = 0.014; insula: R = -0.650, P = 0.003; cingulum: R = -0.665, P = 0.002) except occipital lobe (R = -0.417, P = 0.076). The hypometabolism to [18F]-FDG PET in AD patients also showed negative correlations with regional [18F]-APN-1607 binding in some signature areas of AD (temporal lobe: R = -0.530, P = 0.020; parietal lobe: R = -0.637, P = 0.003; occipital lobe: R = -0.567, P = 0.011). In conclusion, our results suggested that [18F]-APN-1607 PET sensitively detected tau deposition in AD and that individual tauopathy correlated with impaired cerebral glucose metabolism and cognitive function.

Prediction of conversion to Alzheimer's disease using deep survival analysis of MRI images

The prediction of the conversion of healthy individuals and those with mild cognitive impairment to the status of active Alzheimer’s disease is a challenging task. Recently, a survival analysis based upon deep learning was developed to enable predictions regarding the timing of an event in a dataset containing censored data. Here, we investigated whether a deep survival analysis could similarly predict the conversion to Alzheimer’s disease. We selected individuals with mild cognitive impairment and cognitively normal subjects and used the grey matter volumes of brain regions in these subjects as predictive features. We then compared the prediction performances of the traditional standard Cox proportional-hazard model, the DeepHit model and our deep survival model based on a Weibull distribution. Our model achieved a maximum concordance index of 0.835, which was higher than that yielded by the Cox model and comparable to that of the DeepHit model. To our best knowledge, this is the first report to describe the application of a deep survival model to brain magnetic resonance imaging data. Our results demonstrate that this type of analysis could successfully predict the time of an individual’s conversion to Alzheimer’s disease.

Individual brain metabolic connectome indicator based on Kullback-Leibler Divergence Similarity Estimation predicts progression from mild cognitive impairment to Alzheimer's dementia

Purpose

Positron emission tomography (PET) with ¹⁸F-fluorodeoxyglucose (FDG) reveals altered cerebral metabolism in individuals with mild cognitive impairment (MCI) and Alzheimer’s dementia (AD). Previous metabolic connectome analyses derive from groups of patients but do not support the prediction of an individual’s risk of conversion from present MCI to AD. We now present an individual metabolic connectome method, namely the Kullback-Leibler Divergence Similarity Estimation (KLSE), to characterize brain-wide metabolic networks that predict an individual’s risk of conversion from MCI to AD.

Methods

FDG-PET data consisting of 50 healthy controls, 332 patients with stable MCI, 178 MCI patients progressing to AD, and 50 AD patients were recruited from ADNI database. Each individual’s metabolic brain network was ascertained using the KLSE method. We compared intra- and intergroup similarity and difference between the KLSE matrix and group-level matrix, and then evaluated the network stability and inter-individual variation of KLSE. The multivariate Cox proportional hazards model and Harrell’s concordance index (C-index) were employed to assess the prediction performance of KLSE and other clinical characteristics.

Results

The KLSE method captures more pathological connectivity in the parietal and temporal lobes relative to the typical group-level method, and yields detailed individual information, while possessing greater stability of network organization (within-group similarity coefficient, 0.789 for sMCI and 0.731 for pMCI). Metabolic connectome expression was a superior predictor of conversion than were other clinical assessments (hazard ratio (HR) = 3.55; 95% CI, 2.77–4.55; P < 0.001). The predictive performance improved further upon combining clinical variables in the Cox model, i.e., C-indices 0.728 (clinical), 0.730 (group-level pattern model), 0.750 (imaging connectome), and 0.794 (the combined model).

Conclusion

The KLSE indicator identifies abnormal brain networks predicting an individual’s risk of conversion from MCI to AD, thus potentially constituting a clinically applicable imaging biomarker.

Electronic supplementary material

The online version of this article (10.1007/s00259-020-04814-x) contains supplementary material, which is available to authorized users.

Resistance, vulnerability and resilience: A review of the cognitive cerebellum in aging and neurodegenerative diseases

In the context of neurodegeneration and aging, the cerebellum is an enigma. Genetic markers of cellular aging in cerebellum accumulate more slowly than in the rest of the brain, and it generates unknown factors that may slow or even reverse neurodegenerative pathology in animal models of Alzheimer's Disease (AD). Cerebellum shows increased activity in early AD and Parkinson's disease (PD), suggesting a compensatory function that may mitigate early symptoms of neurodegenerative pathophysiology. Perhaps most notably, different parts of the brain accumulate neuropathological markers of AD in a recognized progression and generally, cerebellum is the last brain region to do so. Taken together, these data suggest that cerebellum may be resistant to certain neurodegenerative mechanisms. On the other hand, in some contexts of accelerated neurodegeneration, such as that seen in chronic traumatic encephalopathy (CTE) following repeated traumatic brain injury (TBI), the cerebellum appears to be one of the most susceptible brain regions to injury and one of the first to exhibit signs of pathology. Cerebellar pathology in neurodegenerative disorders is strongly associated with cognitive dysfunction. In neurodegenerative or neurological disorders associated with cerebellar pathology, such as spinocerebellar ataxia, cerebellar cortical atrophy, and essential tremor, rates of cognitive dysfunction, dementia and neuropsychiatric symptoms increase. When the cerebellum shows AD pathology, such as in familial AD, it is associated with earlier onset and greater severity of disease. These data suggest that when neurodegenerative processes are active in the cerebellum, it may contribute to pathological behavioral outcomes. The cerebellum is well known for comparing internal representations of information with observed outcomes and providing real-time feedback to cortical regions, a critical function that is disturbed in neuropsychiatric disorders such as intellectual disability, schizophrenia, dementia, and autism, and required for cognitive domains such as working memory. While cerebellum has reciprocal connections with non-motor brain regions and likely plays a role in complex, goal-directed behaviors, it has proven difficult to establish what it does mechanistically to modulate these behaviors. Due to this lack of understanding, it's not surprising to see the cerebellum reflexively dismissed or even ignored in basic and translational neuropsychiatric literature. The overarching goals of this review are to answer the following questions from primary literature: When the cerebellum is affected by pathology, is it associated with decreased cognitive function? When it is intact, does it play a compensatory or protective role in maintaining cognitive function? Are there theoretical frameworks for understanding the role of cerebellum in cognition, and perhaps, illnesses characterized by cognitive dysfunction? Understanding the role of the cognitive cerebellum in neurodegenerative diseases has the potential to offer insight into origins of cognitive deficits in other neuropsychiatric disorders, which are often underappreciated, poorly understood, and not often treated.

Predictive Value of 18F-Florbetapir and 18F-FDG PET for Conversion from Mild Cognitive Impairment to Alzheimer Dementia

The present study examined the predictive values of amyloid PET, ¹⁸F-FDG PET, and nonimaging predictors (alone and in combination) for development of Alzheimer dementia (AD) in a large population of patients with mild cognitive impairment (MCI). Methods: The study included 319 patients with MCI from the Alzheimer Disease Neuroimaging Initiative database. In a derivation dataset (n = 159), the following Cox proportional-hazards models were constructed, each adjusted for age and sex: amyloid PET using ¹⁸F-florbetapir (pattern expression score of an amyloid-β AD conversion-related pattern, constructed by principle-components analysis); ¹⁸F-FDG PET (pattern expression score of a previously defined ¹⁸F-FDG-based AD conversion-related pattern, constructed by principle-components analysis); nonimaging (functional activities questionnaire, apolipoprotein E, and mini-mental state examination score); ¹⁸F-FDG PET + amyloid PET; amyloid PET + nonimaging; ¹⁸F-FDG PET + nonimaging; and amyloid PET + ¹⁸F-FDG PET + nonimaging. In a second step, the results of Cox regressions were applied to a validation dataset (n = 160) to stratify subjects according to the predicted conversion risk. Results: On the basis of the independent validation dataset, the ¹⁸F-FDG PET model yielded a significantly higher predictive value than the amyloid PET model. However, both were inferior to the nonimaging model and were significantly improved by the addition of nonimaging variables. The best prediction accuracy was reached by combining ¹⁸F-FDG PET, amyloid PET, and nonimaging variables. The combined model yielded 5-y free-of-conversion rates of 100%, 64%, and 24% for the low-, medium- and high-risk groups, respectively. Conclusion: ¹⁸F-FDG PET, amyloid PET, and nonimaging variables represent complementary predictors of conversion from MCI to AD. Especially in combination, they enable an accurate stratification of patients according to their conversion risks, which is of great interest for patient care and clinical trials.

Controls-based denoising, a new approach for medical image analysis, improves prediction of conversion to Alzheimer's disease with FDG-PET

OBJECTIVE

The pattern expression score (PES), i.e., the degree to which a pathology-related pattern is present, is frequently used in FDG-brain-PET analysis and has been shown to be a powerful predictor of conversion to Alzheimer's disease (AD) in mild cognitive impairment (MCI). Since, inevitably, the PES is affected by non-pathological variability, our aim was to improve classification with the simple, yet novel approach to identify patterns of non-pathological variance in a separate control sample using principal component analysis and removing them from patient data (controls-based denoising, CODE) before calculating the PES.

METHODS

Multi-center FDG-PET from 220 MCI patients (64 non-converter, follow-up ≥ 4 years; 156 AD converter, time-to-conversion ≤ 4 years) were obtained from the ADNI database. Patterns of non-pathological variance were determined from 262 healthy controls. An AD pattern was calculated from AD patients and controls. We predicted AD conversion based on PES only and on PES combined with neuropsychological features and ApoE4 genotype. We compared classification performance achieved with and without CODE and with a standard machine learning approach (support vector machine).

RESULTS

Our model predicts that CODE improves the signal-to-noise ratio of AD-PES by a factor of 1.5. PES-based prediction of AD conversion improved from AUC 0.80 to 0.88 (p= 0.001, DeLong's method), sensitivity 69 to 83%, specificity 81% to 88% and Matthews correlation coefficient (MCC) 0.45 to 0.66. Best classification (0.93 AUC) was obtained when combining the denoised PES with clinical features.

CONCLUSIONS

CODE, applied in its basic form, significantly improved prediction of conversion based on PES. The achieved classification performance was higher than with a standard machine learning algorithm, which was trained on patients, explainable by the fact that CODE used additional information (large sample of healthy controls). We conclude that the proposed, novel method is a powerful tool for improving medical image analysis that offers a wide spectrum of biomedical applications, even beyond image analysis.