Discriminating Aging Cognitive Decline Spectrum Using PET and Magnetic Resonance Image Features.

BACKGROUND: The population aging increased the prevalence of brain diseases, like Alzheimer's disease (AD). Early identification of individuals with higher odds of cognitive decline is essential to maintain quality of life. Imaging evaluation of individuals at risk of cognitive decline includes biomarkers extracted from brain positron emission tomography (PET) and structural magnetic resonance imaging (MRI). OBJECTIVE: We propose investigating ensemble models to classify groups in the aging cognitive decline spectrum by combining features extracted from single imaging modalities and combinations of imaging modalities (FDG+AMY+MRI, and a PET ensemble). METHODS: We group imaging data of 131 individuals into four classes related to the individuals' cognitive assessment in baseline and follow-up: stable cognitive non-impaired; individuals converting to mild cognitive impairment (MCI) syndrome; stable MCI; and Alzheimer's clinical syndrome. We assess the performance of four algorithms using leave-one-out cross-validation: decision tree classifier, random forest (RF), light gradient boosting machine (LGBM), and categorical boosting (CAT). The performance analysis of models is evaluated using balanced accuracy before and after using Shapley Additive exPlanations with recursive feature elimination (SHAP-RFECV) method. RESULTS: Our results show that feature selection with CAT or RF algorithms have the best overall performance in discriminating early cognitive decline spectrum mainly using MRI imaging features. CONCLUSION: Use of CAT or RF algorithms with SHAP-RFECV shows good discrimination of early stages of aging cognitive decline, mainly using MRI image features. Further work is required to analyze the impact of selected brain regions and their correlation with cognitive decline spectrum.

Innate immune cells and myelin profile in multiple sclerosis: a multi-tracer PET/MR study.

PURPOSE: Neuropathological studies have demonstrated distinct profiles of microglia activation and myelin injury among different multiple sclerosis (MS) phenotypes and disability stages. PET imaging using specific tracers may uncover the in vivo molecular pathology and broaden the understanding of the disease heterogeneity. METHODS: We used the 18-kDa translocator protein (TSPO) tracer (R)-[11C]PK11195 and [11C]PIB PET images acquired in a hybrid PET/MR 3 T system to characterize, respectively, the profile of innate immune cells and myelin content in 47 patients with MS compared to 18 healthy controls (HC). For the volume of interest (VOI)-based analysis of the dynamic data, (R)-[11C]PK11195 distribution volume (VT) was determined for each subject using a metabolite-corrected arterial plasma input function while [11C]PIB distribution volume ratio (DVR) was estimated using a reference region extracted by a supervised clustering algorithm. A voxel-based analysis was also performed using Statistical Parametric Mapping. Functional disability was evaluated by the Expanded Disability Status Scale (EDSS), Multiple Sclerosis Functional Composite (MSFC), and Symbol Digit Modality Test (SDMT). RESULTS: In the VOI-based analysis, [11C]PIB DVR differed between patients and HC in the corpus callosum (P = 0.019) while no differences in (R)-[11C]PK11195 VT were observed in patients relative to HC. Furthermore, no correlations or associations were observed between both tracers within the VOI analyzed. In the voxel-based analysis, high (R)-[11C]PK11195 uptake was observed diffusively in the white matter (WM) when comparing the progressive phenotype and HC, and lower [11C]PIB uptake was observed in certain WM regions when comparing the relapsing-remitting phenotype and HC. None of the tracers were able to differentiate phenotypes at voxel or VOI level in our cohort. Linear regression models adjusted for age, sex, and phenotype demonstrated that higher EDSS was associated with an increased (R)-[11C]PK11195 VT and lower [11C]PIB DVR in corpus callosum (P = 0.001; P = 0.023), caudate (P = 0.015; P = 0.008), and total T2 lesion (P = 0.007; P = 0.012), while better cognitive scores in SDMT were associated with higher [11C]PIB DVR in the corpus callosum (P = 0.001), and lower (R)-[11C]PK11195 VT (P = 0.013). CONCLUSIONS: Widespread innate immune cells profile and marked loss of myelin in T2 lesions and regions close to the ventricles may occur independently and are associated with disability, in both WM and GM structures.

Cognitive Decline Assessment: A Review From Medical Imaging Perspective.

Aging is a complex process that involves changes at both molecular and morphological levels. However, our understanding of how aging affects brain anatomy and function is still poor. In addition, numerous biomarkers and imaging markers, usually associated with neurodegenerative diseases such as Alzheimer's disease (AD), have been clinically used to study cognitive decline. However, the path of cognitive decline from healthy aging to a mild cognitive impairment (MCI) stage has been studied only marginally. This review presents aspects of cognitive decline assessment based on the imaging differences between individuals cognitively unimpaired and in the decline spectrum. Furthermore, we discuss the relationship between imaging markers and the change in their patterns with aging by using neuropsychological tests. Our goal is to delineate how aging has been studied by using medical imaging tools and further explore the aging brain and cognitive decline. We find no consensus among the biomarkers to assess the cognitive decline and its relationship with the cognitive decline trajectory. Brain glucose hypometabolism was found to be directly related to aging and indirectly to cognitive decline. We still need to understand how to quantify an expected hypometabolism during cognitive decline during aging. The Aß burden should be longitudinally studied to achieve a better consensus on its association with changes in the brain and cognition decline with aging. There exists a lack of standardization of imaging markers that highlight the need for their further improvement. In conclusion, we argue that there is a lot to investigate and understand cognitive decline better and seek a window for a suitable and effective treatment strategy.

Volume-of-interest-based supervised cluster analysis for pseudo-reference region selection in [18F]DPA-714 PET imaging of the rat brain.

METHOD: Aim of this study was to automatically select a suitable pseudo-reference brain region for the accurate, non-invasive quantification of neuroinflammation in a rat brain using dynamic [18F]DPA-714 PET imaging. PROCEDURES: A supervised clustering analysis approach considering three kinetic classes (SVCA3) was used to select an appropriate pseudo-reference brain region. This pseudo-reference region was determined by selecting only brain regions with low specific tracer uptake (SVCA3low) or by taking into account all brain regions and weighting each brain region with the corresponding fraction of low specific binding (SVCA3wlow). Both SVCA3 approaches were evaluated in an animal model of neuro-inflammation induced by lipopolysaccharide injection in the right striatum of female Wistar rats. For this study setup, a population of 25 female Wistar rats received a dynamic PET scan after injection of ~ 60 MBq [18F]DPA-714. Animals were scanned at baseline (n = 3) and at different time points after inducing neuroinflammation: 1 day (n = 3), 3 days (n = 12), 7 days (n = 4), and 30 days (n = 3). Binding potential (BP) values using a simplified reference tissue model (SRTM) and the contralateral striatum as pseudo-reference region were considered as a reference method (BPL STR) and compared with SRTM BP values using a pseudo-reference region obtained by either the SVCA3low or SVCA3wlow approach for both a 90- and 120-min acquisition time interval. RESULTS: For the right striatum, SRTM BP values using a SVCA3low- or SVCA3wlow-based pseudo-reference region demonstrated a strong and highly significant correlation with SRTM BPL STR values (Spearman r ≥ 0.89, p < 0.001). For the SVCA3low approach, Friedman tests revealed no significant difference with SRTM BPL STR values for a 120-min acquisition time while small but signification differences were found for a 90-min acquisition time (p < 0.05). For the SVCA3wlow approach, highly signification differences (p < 0.001) were found with SRTM BPL STR values for both a 90- and 120-min acquisition time interval. CONCLUSIONS: A SVCA3 approach using three kinetic classes allowed the automatic selection of pseudo-reference brain regions with low specific tracer binding for accurate and non-invasive quantification of rat brain PET imaging using [18F]DPA-714. A shorter acquisition time interval of 90 min can be considered with only limited impact on the SVCA3-based selection of the pseudo-reference brain regions.