There is no gold standard for the diagnosis of Alzheimer’s disease (AD), except for autopsies, which motivates the use of unsupervised learning. A mixture of regressions is an unsupervised method that can simultaneously identify clusters from multiple biomarkers while learning within-cluster demographic effects. Cerebrospinal fluid (CSF) biomarkers for AD have detection limits, which create additional challenges. We apply a mixture of regressions with a multivariate truncated Gaussian distribution (also called a censored multivariate Gaussian mixture of regressions or a mixture of multivariate Tobit regressions) to over 3000 participants from the Emory Goizueta Alzheimer’s Disease Research Center and Emory Healthy Brain Study to examine amyloid-beta peptide 1–42 (Abeta42), total tau protein and phosphorylated tau protein in CSF with known detection limits. We address three gaps in the literature on the mixture of regressions with a truncated multivariate Gaussian distribution: software availability; inference; and clustering accuracy. We discovered three clusters that tend to align with an AD group, a normal control profile, and non-AD pathology. The CSF profiles differed by race, gender, and the genetic marker ApoE4, highlighting the importance of considering demographic factors in unsupervised learning with detection limits. Notably, African American participants in the AD-like group had significantly lower tau burden.

Multi-channel phased receive arrays have been widely adopted for magnetic resonance imaging (MRI) and spectroscopy (MRS). An important step in the use of receive arrays for MRS is the combination of spectra collected from individual coil channels. The goal of this work was to implement an improved strategy termed OpTIMUS (i.e., optimized truncation to integrate multi-channel MRS data using rank-R singular value decomposition) for combining data from individual channels. OpTIMUS relies on spectral windowing coupled with a rank-R decomposition to calculate the optimal coil channel weights. MRS data acquired from a brain spectroscopy phantom and 11 healthy volunteers were first processed using a whitening transformation to remove correlated noise. Whitened spectra were then iteratively windowed or truncated, followed by a rank-R singular value decomposition (SVD) to empirically determine the coil channel weights. Spectra combined using the vendor-supplied method, signal/noise2 weighting, previously reported whitened SVD (rank-1), and OpTIMUS were evaluated using the signal-to-noise ratio (SNR). Significant increases in SNR ranging from 6% to 33% (P ≤ 0.05) were observed for brain MRS data combined with OpTIMUS compared with the three other combination algorithms. The assumption that a rank-1 SVD maximizes SNR was tested empirically, and a higher rank-R decomposition, combined with spectral windowing prior to SVD, resulted in increased SNR.

Two controversial tenets of metapopulation biology are whether patch quality and the surrounding matrix are more important to turnover (colonisation and extinction) than biogeography (patch area and isolation) and whether factors governing turnover during equilibrium also dominate nonequilibrium dynamics. We tested both tenets using 18 years of surveys for two secretive wetland birds, black and Virginia rails, during (1) a period of equilibrium with stable occupancy and (2) after drought and arrival of West Nile Virus (WNV), which resulted in WNV infections in rails, increased extinction and decreased colonisation probabilities modified by WNV, nonequilibrium dynamics for both species and occupancy decline for black rails. Area (primarily) and isolation (secondarily) drove turnover during both stable and unstable metapopulation dynamics, greatly exceeding the effects of patch quality and matrix conditions. Moreover, slopes between turnover and patch characteristics changed little between equilibrium and nonequilibrium, confirming the overriding influences of biogeographic factors on turnover.

Brain temperature is an understudied parameter relevant to brain injury and ischemia. To advance our understanding of thermal dynamics in the human brain, combined with the challenges of routine experimental measurements, a biophysical modeling framework was developed to facilitate individualized brain temperature predictions. Model-predicted brain temperatures using our fully conserved model were compared with whole brain chemical shift thermometry acquired in 30 healthy human subjects (15 male and 15 female, age range 18–36 years old). Magnetic resonance (MR) thermometry, as well as structural imaging, angiography, and venography, were acquired prospectively on a Siemens Prisma whole body 3 T MR scanner. Bland–Altman plots demonstrate agreement between model-predicted and MR-measured brain temperatures at the voxel-level. Regional variations were similar between predicted and measured temperatures (< 0.55 °C for all 10 cortical and 12 subcortical regions of interest), and subcortical white matter temperatures were higher than cortical regions. We anticipate the advancement of brain temperature as a marker of health and injury will be facilitated by a well-validated computational model which can enable predictions when experiments are not feasible.

While hippocampal atrophy and its regional susceptibility to Alzheimer’s disease (AD) are well reported at late stages of AD, studies of the asymptomatic stage of AD are limited but could elucidate early stage pathophysiology as well as provide predictive biomarkers. In this study, we performed multi-modal magnetic resonance imaging (MRI) to estimate morphometry, functional connectivity, and tissue microstructure of hippocampal subfields in cognitively normal adults including those with asymptomatic AD. High-resolution resting-state functional, diffusion and structural MRI, cerebral spinal fluid (CSF), and neuropsychological evaluations were performed in healthy young adults (HY: n = 40) and healthy older adults with negative (HO−: n = 47) and positive (HO+ : n = 25) CSF biomarkers of AD. Morphometry, functional connectivity, and tissue microstructure were estimated from the structural, functional, and diffusion MRI images, respectively. Our results indicated that normal aging affected morphometry, connectivity, and microstructure in all hippocampal subfields, while the subiculum and CA1-3 demonstrated the greatest sensitivity to asymptomatic AD pathology. Tau, rather than amyloid-β, was closely associated with imaging-derived synaptic and microstructural measures. Microstructural metrics were significantly associated with neuropsychological assessments. These findings suggest that the subiculum and CA1-3 are the most vulnerable in asymptomatic AD and tau level is driving these early changes.

The exclusion of high-motion participants can reduce the impact of motion in functional Magnetic Resonance Imaging (fMRI) data. However, the exclusion of high-motion participants may change the distribution of clinically relevant variables in the study sample, and the resulting sample may not be representative of the population. Our goals are two-fold: 1) to document the biases introduced by common motion exclusion practices in functional connectivity research and 2) to introduce a framework to address these biases by treating excluded scans as a missing data problem. We use a study of autism spectrum disorder in children without an intellectual disability to illustrate the problem and the potential solution. We aggregated data from 545 children (8–13 years old) who participated in resting-state fMRI studies at Kennedy Krieger Institute (173 autistic and 372 typically developing) between 2007 and 2020. We found that autistic children were more likely to be excluded than typically developing children, with 28.5% and 16.1% of autistic and typically developing children excluded, respectively, using a lenient criterion and 81.0% and 60.1% with a stricter criterion. The resulting sample of autistic children with usable data tended to be older, have milder social deficits, better motor control, and higher intellectual ability than the original sample. These measures were also related to functional connectivity strength among children with usable data. This suggests that the generalizability of previous studies reporting naïve analyses (i.e., based only on participants with usable data) may be limited by the selection of older children with less severe clinical profiles because these children are better able to remain still during an rs-fMRI scan. We adapt doubly robust targeted minimum loss based estimation with an ensemble of machine learning algorithms to address these data losses and the resulting biases. The proposed approach selects more edges that differ in functional connectivity between autistic and typically developing children than the naïve approach, supporting this as a promising solution to improve the study of heterogeneous populations in which motion is common.

Joint and Individual Variation Explained (JIVE) is a model that decomposes multiple datasets obtained on the same subjects into shared structure, structure unique to each dataset, and noise. JIVE is an important tool for multimodal data integration in neuroimaging. The two most common algorithms are R.JIVE, an iterative approach, and AJIVE, which uses principal angle analysis. The joint structure in JIVE is defined by shared subspaces, but interpreting these subspaces can be challenging. In this paper, we reinterpret AJIVE as a canonical correlation analysis of principal component scores. This reformulation, which we call CJIVE, (1) provides an intuitive view of AJIVE; (2) uses a permutation test for the number of joint components; (3) can be used to predict subject scores for out-of-sample observations; and (4) is computationally fast. We conduct simulation studies that show CJIVE and AJIVE are accurate when the total signal ranks are correctly specified but, generally inaccurate when the total ranks are too large. CJIVE and AJIVE can still extract joint signal even when the joint signal variance is relatively small. JIVE methods are applied to integrate functional connectivity (resting-state fMRI) and structural connectivity (diffusion MRI) from the Human Connectome Project. Surprisingly, the edges with largest loadings in the joint component in functional connectivity do not coincide with the same edges in the structural connectivity, indicating more complex patterns than assumed in spatial priors. Using these loadings, we accurately predict joint subject scores in new participants. We also find joint scores are associated with fluid intelligence, highlighting the potential for JIVE to reveal important shared structure.

Purpose: The presence of orientation-dependent susceptibility artifacts in magnetic resonance chemical shift thermometry (CST) can confound accurate temperature calculations. Here, we quantify the effect of white matter (WM) tract orientation on CST due to tissue-specific susceptibility. Methods: Twenty-nine healthy volunteers (27 ± 4 years old) were scanned on a 3 T MR scanner with a 32-channel head coil. Diffusion tensor imaging (DTI), T1-weighted imaging, and single voxel spectroscopy (SVS) for CST were acquired. Participants were then asked to rotate their head ∼3–5° (yaw or roll) to alter the orientation of WM tracts relative to the external magnetic field. After head rotation, a second SVS scan and T1-weighted imaging were acquired. The WM-fraction-normalized DTI principal eigenvector (V1) images were used to calculate the length of the x-y component of V1, which was used as a surrogate for WM tracts perpendicular to B0. A linear regression model was used to determine the relationship between the perpendicular WM tracts and brain temperature. Results: Significant temperature differences between post- and pre-head rotation scans were observed for brain (−0.72 °C ± 1.36 °C, p = 0.01) but not body (0.012 °C ± 0.07 °C, p = 0.37) temperatures. The difference in brain temperature was positively associated with the corresponding change in perpendicular WM tracts after head rotation (R2 = 0.26, p = 0.005). Conclusion: Our results indicate WM tract orientation affects temperature calculations, suggesting artifacts from orientation-dependent susceptibility may be present in CST.

Independent component analysis (ICA) is an unsupervised learning method popular in functional magnetic resonance imaging (fMRI). Group ICA has been used to search for biomarkers in neurological disorders including autism spectrum disorder and dementia. However, current methods use a principal component analysis (PCA) step that may remove low-variance features. Linear non-Gaussian component analysis (LNGCA) enables simultaneous dimension reduction and feature estimation including low-variance features in single-subject fMRI. A group LNGCA model is proposed to extract group components shared by more than one subject. Unlike group ICA methods, this novel approach also estimates individual (subject-specific) components orthogonal to the group components. To determine the total number of components in each subject, a parametric resampling test is proposed that samples spatially correlated Gaussian noise to match the spatial dependence observed in data. In simulations, estimated group components achieve higher accuracy compared to group ICA. The method is applied to a resting-state fMRI study on autism spectrum disorder in 342 children (252 typically developing, 90 with autism), where the group signals include resting-state networks. The discovered group components appear to exhibit different levels of temporal engagement in autism versus typically developing children, as revealed using group LNGCA. This novel approach to matrix decomposition is a promising direction for feature detection in neuroimaging.

Environmental fecal contamination is common in many low-income cities, contributing to a high burden of enteric infections and associated negative sequelae. To evaluate the impact of a shared onsite sanitation intervention in Maputo, Mozambique on enteric pathogens in the domestic environment, we collected 179 soil samples at shared latrine entrances from intervention (n = 49) and control (n = 51) compounds during baseline (preintervention) and after 24 months (postintervention) as part of the Maputo Sanitation Trial. We tested soils for the presence of nucleic acids associated with 18 enteric pathogens using a multiplex reverse transcription qPCR platform. We detected at least one pathogen-associated gene target in 91% (163/179) of soils and a median of 3 (IQR = 1, 5) pathogens. Using a difference-in-difference analysis and adjusting for compound population, visibly wet soil, sun exposure, wealth, temperature, animal presence, and visible feces, we estimate the intervention reduced the probability of detecting ≥1 pathogen gene by 15% (adjusted prevalence ratio, aPR = 0.85; 95% CI: 0.70, 1.0) and the total number of pathogens by 35% (aPR = 0.65; 0.44, 0.95) in soil 24 months following the intervention. These results suggest that the intervention reduced the presence of some fecal contamination in the domestic environment, but pathogen detection remained prevalent 24 months following the introduction of new latrines.