Explainable AI-based Deep-SHAP for mapping the multivariate relationships between regional neuroimaging biomarkers and cognition.

BACKGROUND: Mild cognitive impairment (MCI)/Alzheimer's disease (AD) is associated with cognitive decline beyond normal aging and linked to the alterations of brain volume quantified by magnetic resonance imaging (MRI) and amyloid-beta (Aß) quantified by positron emission tomography (PET). Yet, the complex relationships between these regional imaging measures and cognition in MCI/AD remain unclear. Explainable artificial intelligence (AI) may uncover such relationships. METHOD: We integrate the AI-based deep learning neural network and Shapley additive explanations (SHAP) approaches and introduce the Deep-SHAP method to investigate the multivariate relationships between regional imaging measures and cognition. After validating this approach on simulated data, we apply it to real experimental data from MCI/AD patients. RESULTS: Deep-SHAP significantly predicted cognition using simulated regional features and identified the ground-truth simulated regions as the most significant multivariate predictors. When applied to experimental MRI data, Deep-SHAP revealed that the insula, lateral occipital, medial frontal, temporal pole, and occipital fusiform gyrus are the primary contributors to global cognitive decline in MCI/AD. Furthermore, when applied to experimental amyloid Pittsburgh compound B (PiB)-PET data, Deep-SHAP identified the key brain regions for global cognitive decline in MCI/AD as the inferior temporal, parahippocampal, inferior frontal, supratemporal, and lateral frontal gray matter. CONCLUSION: Deep-SHAP method uncovered the multivariate relationships between regional brain features and cognition, offering insights into the most critical modality-specific brain regions involved in MCI/AD mechanisms.

Application of glutamate weighted CEST in brain imaging of nicotine dependent participants in vivo at 7T.

INTRODUCTION: With nicotine dependence being a significant healthcare issue worldwide there is a growing interest in developing novel therapies and diagnostic aids to assist in treating nicotine addiction. Glutamate (Glu) plays an important role in cognitive function regulation in a wide range of conditions including traumatic brain injury, aging, and addiction. Chemical exchange saturation transfer (CEST) imaging via ultra-high field MRI can image the exchange of certain saturated labile protons with the surrounding bulk water pool, making the technique a novel tool to investigate glutamate in the context of addiction. The aim of this work was to apply glutamate weighted CEST (GluCEST) imaging to study the dorsal anterior cingulate cortex (dACC) in a small population of smokers and non-smokers to determine its effectiveness as a biomarker of nicotine use. METHODS: 2D GluCEST images were acquired on 20 healthy participants: 10 smokers (ages 29-50) and 10 non-smokers (ages 25-69), using a 7T MRI system. T1-weighted images were used to segment the GluCEST images into white and gray matter tissue and further into seven gray matter regions. Wilcoxon rank-sum tests were performed, comparing mean GluCEST contrast between smokers and non-smokers across brain regions. RESULTS: GluCEST levels were similar between smokers and non-smokers; however, there was a moderate negative age dependence (R2 = 0.531) in smokers within the cingulate gyrus. CONCLUSION: Feasibility of GluCEST imaging was demonstrated for in vivo investigation of smokers and non-smokers to assess glutamate contrast differences as a potential biomarker with a moderate negative age correlation in the cingulate gyrus suggesting reward network involvement.

Dysregulated Salience Network Control over Default-Mode and Central-Executive Networks in Schizophrenia Revealed Using Stochastic Dynamical Causal Modeling.

Introduction: Neuroimaging studies suggest that the human brain consists of intrinsically organized, large-scale neural networks. Among these networks, the interplay among the default-mode network (DMN), salience network (SN), and central-executive network (CEN) has been widely used to understand the functional interaction patterns in health and disease. This triple network model suggests that the SN causally controls over the DMN and CEN in healthy individuals. This interaction is often referred to as SN's dynamic regulating mechanism. However, such interactions are not well understood in individuals with schizophrenia. Methods: In this study, we leveraged resting-state functional magnetic resonance imaging data from schizophrenia (n = 67) and healthy controls (n = 81) and evaluated the directional functional interactions among DMN, SN, and CEN using stochastic dynamical causal modeling methodology. Results: In healthy controls, our analyses replicated previous findings that SN regulates DMN and CEN activities (Mann-Whitney U test; p < 10-8). In schizophrenia, however, our analyses revealed a disrupted SN-based controlling mechanism over the DMN and CEN (Mann-Whitney U test; p < 10-16). Conclusions: These results indicate that the disrupted controlling mechanism of SN over the other two neural networks may be a candidate neuroimaging phenotype in schizophrenia.

Predicting cognitive dysfunction and regional hubs using Braak staging amyloid-beta biomarkers and machine learning.

Mild cognitive impairment (MCI) is a transitional stage between normal aging and early Alzheimer's disease (AD). The presence of extracellular amyloid-beta (Aß) in Braak regions suggests a connection with cognitive dysfunction in MCI/AD. Investigating the multivariate predictive relationships between regional Aß biomarkers and cognitive function can aid in the early detection and prevention of AD. We introduced machine learning approaches to estimate cognitive dysfunction from regional Aß biomarkers and identify the Aß-related dominant brain regions involved with cognitive impairment. We employed Aß biomarkers and cognitive measurements from the same individuals to train support vector regression (SVR) and artificial neural network (ANN) models and predict cognitive performance solely based on Aß biomarkers on the test set. To identify Aß-related dominant brain regions involved in cognitive prediction, we built the local interpretable model-agnostic explanations (LIME) model. We found elevated Aß in MCI compared to controls and a stronger correlation between Aß and cognition, particularly in Braak stages III-IV and V-VII (p < 0.05) biomarkers. Both SVR and ANN, especially ANN, showed strong predictive relationships between regional Aß biomarkers and cognitive impairment (p < 0.05). LIME integrated with ANN showed that the parahippocampal gyrus, inferior temporal gyrus, and hippocampus were the most decisive Braak regions for predicting cognitive decline. Consistent with previous findings, this new approach suggests relationships between Aß biomarkers and cognitive impairment. The proposed analytical framework can estimate cognitive impairment from Braak staging Aß biomarkers and delineate the dominant brain regions collectively involved in AD pathophysiology.

In vivo reproducibility of 3D relayed NOE in the healthy human brain at 7 T.

PURPOSE: Nuclear Overhauser effect (NOE) is based on dipolar cross-relaxation mechanism that enables the indirect detection of aliphatic protons via the water proton signal. This work focuses on determining the reproducibility of NOE magnetization transfer ratio (NOEMTR ) and isolated or relayed NOE (rNOE) contributions to the NOE MRI of the healthy human brain at 7 Tesla (T). METHODS: We optimized the B 1 + $$ {\mathrm{B}}_1^{+} $$ amplitude and length of the saturation pulse by acquiring NOE images with different B 1 + $$ {\mathrm{B}}_1^{+} $$ values with multiple saturation lengths. Repeated NOE MRI measurements were made on five healthy volunteers by using optimized saturation pulse parameters including correction of B0 and B 1 + $$ {\mathrm{B}}_1^{+} $$ inhomogeneities. To isolate the individual contributions from z-spectra, we have fit the NOE z-spectra using multiple Lorentzians and calculated the total contribution from each pool contributing to the overall NOEMTR contrast. RESULTS: We found that a saturation amplitude of 0.72 µT and a length of 3 s provided the highest contrast. We found that the mean NOEMTR value in gray matter (GM) was 26%, and in white matter (WM) was 33.3% across the 3D slab of the brain. The mean rNOE contributions from GM and WM values were 8.9% and 9.6%, which were ∼10% of the corresponding total NOEMTR signal. The intersubject coefficient of variations (CoVs) of NOEMTR from GM and WM were 4.5% and 6.5%, respectively, whereas the CoVs of rNOE were 4.8% and 5.6%, respectively. The intrasubject CoVs of the NOEMTR range was 2.1%-4.2%, and rNOE range was 2.9%-10.5%. CONCLUSION: This work has demonstrated an excellent reproducibility of both inter- and intrasubject NOEMTR and rNOE metrics in healthy human brains at 7 T.

Integrating 1H MRS and deuterium labeled glucose for mapping the dynamics of neural metabolism in humans.

In the technique presented here, dubbed 'qMRS', we quantify the change in 1H MRS signal following administration of 2H-labeled glucose. As in recent human DMRS studies, we administer [6,6'-2H2]-glucose orally to healthy subjects. Since 2H is not detectable by 1H MRS, the transfer of the 2H label from glucose to a downstream metabolite leads to a reduction in the corresponding 1H MRS resonance of the metabolite, even if the total concentration of both isoforms remains constant. Moreover, introduction of the deuterium label alters the splitting pattern of the proton resonances, making indirect detection of the deuterated forms- as well as the direct detection of the decrease in unlabeled form- possible even without a 2H coil. Because qMRS requires only standard 1H MRS acquisition methods, it can be performed using commonly implemented single voxel spectroscopy (SVS) and chemical shift imaging (CSI) sequences. In this work, we implement qMRS in semi-LASER based CSI, generating dynamic maps arising from the fitted spectra, and demonstrating the feasibility of using qMRS and qCSI to monitor dynamic metabolism in the human brain using a 7T scanner with no auxiliary hardware.

Salience network anatomical and molecular markers are linked with cognitive dysfunction in mild cognitive impairment.

BACKGROUND AND PURPOSE: Recent studies indicate disrupted functional mechanisms of salience network (SN) regions-right anterior insula, left anterior insula, and anterior cingulate cortex-in mild cognitive impairment (MCI). However, the underlying anatomical and molecular mechanisms in these regions are not clearly understood yet. It is also unknown whether integration of multimodal-anatomical and molecular-markers could predict cognitive impairment better in MCI. METHODS: Herein we quantified anatomical volumetric markers via structural MRI and molecular amyloid markers via PET with Pittsburgh compound B in SN regions of MCI (n = 33) and healthy controls (n = 27). From these markers, we built support vector machine learning models aiming to estimate cognitive dysfunction in MCI. RESULTS: We found that anatomical markers are significantly reduced and molecular markers are significantly elevated in SN nodes of MCI compared to healthy controls (p < .05). These altered markers in MCI patients were associated with their worse cognitive performance (p < .05). Our machine learning-based modeling further suggested that the integration of multimodal markers predicts cognitive impairment in MCI superiorly compared to using single modality-specific markers. CONCLUSIONS: These findings shed light on the underlying anatomical volumetric and molecular amyloid alterations in SN regions and show the significance of multimodal markers integration approach in better predicting cognitive impairment in MCI.

Multimodality assessment of heart failure with preserved ejection fraction skeletal muscle reveals differences in the machinery of energy fuel metabolism.

AIMS: Skeletal muscle (SkM) abnormalities may impact exercise capacity in patients with heart failure with preserved ejection fraction (HFpEF). We sought to quantify differences in SkM oxidative phosphorylation capacity (OxPhos), fibre composition, and the SkM proteome between HFpEF, hypertensive (HTN), and healthy participants. METHODS AND RESULTS: Fifty-nine subjects (20 healthy, 19 HTN, and 20 HFpEF) performed a maximal-effort cardiopulmonary exercise test to define peak oxygen consumption (VO2, peak ), ventilatory threshold (VT), and VO2 efficiency (ratio of total work performed to O2 consumed). SkM OxPhos was assessed using Creatine Chemical-Exchange Saturation Transfer (CrCEST, n = 51), which quantifies unphosphorylated Cr, before and after plantar flexion exercise. The half-time of Cr recovery (t1/2, Cr ) was taken as a metric of in vivo SkM OxPhos. In a subset of subjects (healthy = 13, HTN = 9, and HFpEF = 12), percutaneous biopsy of the vastus lateralis was performed for myofibre typing, mitochondrial morphology, and proteomic and phosphoproteomic analysis. HFpEF subjects demonstrated lower VO2,peak , VT, and VO2 efficiency than either control group (all P < 0.05). The t1/2, Cr was significantly longer in HFpEF (P = 0.005), indicative of impaired SkM OxPhos, and correlated with cycle ergometry exercise parameters. HFpEF SkM contained fewer Type I myofibres (P = 0.003). Proteomic analyses demonstrated (a) reduced levels of proteins related to OxPhos that correlated with exercise capacity and (b) reduced ERK signalling in HFpEF. CONCLUSIONS: Heart failure with preserved ejection fraction patients demonstrate impaired functional capacity and SkM OxPhos. Reductions in the proportions of Type I myofibres, proteins required for OxPhos, and altered phosphorylation signalling in the SkM may contribute to exercise intolerance in HFpEF.

Recovery kinetics of creatine in mild plantar flexion exercise using 3D creatine CEST imaging at 7 Tesla.

PURPOSE: Two-dimensional creatine CEST (2D-CrCEST), with a slice thickness of 10-20 mm and temporal resolution (τRes ) of about 30 seconds, has previously been shown to capture the creatine-recovery kinetics in healthy controls and in patients with abnormal creatine-kinase kinetics following the mild plantar flexion exercise. Since the distribution of disease burden may vary across the muscle length for many musculoskeletal disorders, there is a need to increase coverage in the slice-encoding direction. Here, we demonstrate the feasibility of 3D-CrCEST with τRes of about 30 seconds, and propose an improved voxel-wise B1+ -calibration approach for CrCEST. METHODS: The current 7T study with enrollment of 5 volunteers involved collecting the baseline CrCEST imaging for the first 2 minutes, followed by 2 minutes of plantar flexion exercise and then 8 minutes of postexercise CrCEST imaging, to detect the temporal evolution of creatine concentration following exercise. RESULTS: Very good repeatability of 3D-CrCEST findings for activated muscle groups on an intraday and interday basis was established, with coefficient of variance of creatine recovery constants (τCr ) being 7%-15.7%, 7.5%, and 5.8% for lateral gastrocnemius, medial gastrocnemius, and peroneus longus, respectively. We also established a good intraday and interday scan repeatability for 3D-CrCEST and also showed good correspondence between τCr measurements using 2D-CrCEST and 3D-CrCEST acquisitions. CONCLUSION: In this study, we demonstrated for the first time the feasibility and the repeatability of the 3D-CrCEST method in calf muscle with improved B1+ correction to measure creatine-recovery kinetics within a large 3D volume of calf muscle.