Probabilistic Independent Component Analysis Research Articles

Interhemispheric functional connectivity: an fMRI study in callosotomized patients.

Functional connectivity (FC) is defined in terms of temporal correlations between physiological signals, which mainly depend upon structural (axonal) connectivity; it is commonly studied using functional magnetic resonance imaging (fMRI). Interhemispheric FC appears mostly supported by the corpus callosum (CC), although several studies investigating this aspect have not provided conclusive evidence. In this context, patients in whom the CC was resected for therapeutic reasons (split-brain patients) provide a unique opportunity for research into this issue. The present study was aimed at investigating with resting-state fMRI the interhemispheric FC in six epileptic patients who have undergone surgical resection of the CC. The analysis was performed using fMRI of the Brain Software Library; the evaluation of interhemispheric FC and the recognition of the resting-state networks (RSNs) were performed using probabilistic independent component analysis. Generally, bilateral brain activation was often observed in primary sensory RSNs, while in the associative areas, such as those composing the default mode and fronto-parietal networks, the activation was often unilateral. These results suggest that even in the absence of the CC, some interhemispheric communication is still present. This residual FC might be supported through extra-callosal pathways that are likely subcortical, making it possible for some interhemispheric integration. Further studies are needed to confirm these conclusions.

A Review of Statistical-Based Fault Detection and Diagnosis with Probabilistic Models

As industrial processes grow increasingly complex, fault identification becomes challenging, and even minor errors can significantly impact both productivity and system safety. Fault detection and diagnosis (FDD) has emerged as a crucial strategy for maintaining system reliability and safety through condition monitoring and abnormality recovery to manage this challenge. Statistical-based FDD methods that rely on large-scale process data and their features have been developed for detecting faults. This paper overviews recent investigations and developments in statistical-based FDD methods, focusing on probabilistic models. The theoretical background of these models is presented, including Bayesian learning and maximum likelihood. We then discuss various techniques and methodologies, e.g., probabilistic principal component analysis (PPCA), probabilistic partial least squares (PPLS), probabilistic independent component analysis (PICA), probabilistic canonical correlation analysis (PCCA), and probabilistic Fisher discriminant analysis (PFDA). Several test statistics are analyzed to evaluate the discussed methods. In industrial processes, these methods require complex matrix operation and cost computational load. Finally, we discuss the current challenges and future trends in FDD.

MRI texture-based radiomics analysis for the identification of altered functional networks in alcoholic patients and animal models.

Alcohol use disorder (AUD) is a complex condition representing a leading risk factor for death, disease and disability. Its high prevalence and severe health consequences make necessary a better understanding of the brain network alterations to improve diagnosis and treatment. The purpose of this study was to evaluate the potential of resting-state fMRI 3D texture features as a novel source of biomarkers to identify AUD brain network alterations following a radiomics approach. A longitudinal study was conducted in Marchigian Sardinian alcohol-preferring msP rats (N=36) who underwent resting-state functional and structural MRI before and after 30 days of alcohol or water consumption. A cross-sectional human study was also conducted among 33 healthy controls and 35 AUD patients. The preprocessed functional data corresponding to control and alcohol conditions were used to perform a probabilistic independent component analysis, identifying seven independent components as resting-state networks. Forty-three radiomic features extracted from each network were compared using a Wilcoxon signed-rank test with Holm correction to identify the network most affected by alcohol consumption. Features extracted from this network were then used in the machine learning process, evaluating two feature selection methods and six predictive models within a nested cross-validation structure. The classification was evaluated by computing the area under the ROC curve. Images were quantized using different numbers of gray-levels to test their influence on the results. The influence of ageing, data preprocessing, and brain iron accumulation were also analyzed. The methodology was validated using structural scans. The striatal network in alcohol-exposed msP rats presented the most significant number of altered features. The radiomics approach supported this result achieving good classification performance in animals (AUC = 0.915±0.100, with 12 features) and humans (AUC = 0.724±0.117, with 9 features) using a random forest model. Using the structural scans, high accuracy was achieved with a multilayer perceptron in both species (animals: AUC > 0.95 with 2 features, humans: AUC > 0.82 with 18 features). The best results were obtained using a feature selection method based on the p-value. The proposed radiomics approach is able to identify AUD patients and alcohol-exposed rats with good accuracy, employing a subset of 3D features extracted from fMRI. Furthermore, it can help identify relevant networks in drug addiction.

Suicidal Ideation Is Associated With Reduced Functional Connectivity and White Matter Integrity in Drug-Naïve Patients With Major Depression.

Depression is a highly prevalent psychiatric disorder affecting millions of people worldwide. Depression is characterized by decreased mood or loss of interest in daily activities, changes in feeding and circadian rhythms and significant impairments in cognitive and executive function. In addition, the occurrence of recurrent thoughts of death and suicidal ideation confers depressed patients a higher risk of suicide than the general population. With this study, we aimed to explore the neural correlates of suicidal ideation in drug-naïve patients diagnosed with depression. Twenty-five patients were scanned using two-different magnetic resonance imaging (MRI) modalities, resting state functional MRI (fMRI) and diffusion tensor imaging (DTI). Resting state allowed the exploration of connectivity patterns in the absence of a specific stimulus and DTI allowed a detailed analysis of structural white matter integrity with measures like fractional anisotropy (FA). Probabilistic independent component analysis (PICA), network-based statistics and tract-based spatial statistics (TBSS) were applied to analyze resting-state fMRI and DTI data, respectively. Our results showed that, in our sample of drug-naïve patients, suicidal ideation was negatively associated with resting-state functional connectivity in the visual networks and with FA in the genu of corpus callosum and in the right anterior corona radiata. In addition, a significant association was identified between suicidal ideation and a functional connectivity network that included connections between regions in the superior and orbitofrontal cortex, the cerebellum, the cingulate gyrus as well as temporal and occipital regions. In conclusion, this work has expanded our knowledge about the possible functional and structural neuronal correlates of suicidal ideation in drug-naïve patients with depression, paving the way for future personalized therapeutic approaches.

Salvage multiple burr hole surgery in patients with Moyamoya disease: efficacy evaluation using probabilistic independent component analysis of dynamic susceptibility contrast perfusion MRI

Multiple burr hole surgery is considered to be an option for achieving indirect revascularization in patients with ischemic Moyamoya disease (MMD). We aimed to investigate the efficacy of stand-alone multiple burr hole surgery for salvage revascularization in patients with MMD by assessing the hemodynamic changes via normalized time-to-peak (nTTP) analysis and independent component analysis (ICA) of preoperative and postoperative dynamic susceptibility contrast (DSC) perfusion MRI data. The DSC perfusion MRI data of 25 hemispheres from 21 patients with MMD, who underwent multiple burr hole surgery for salvage revascularization due to persistent or recurrent symptoms after primary revascularization with modified encephaloduroarteriosynangiosis (mEDAS), were analyzed. The nTTP, which was measured using the region of interests covering the entire surgical hemisphere, was compared between the preoperative and postoperative images. ICA was used to compare the relative arterial and venous components of the surgical hemispheres between the respective preoperative and postoperative images. The median postoperative nTTP (1.80s) was significantly shorter than the median preoperative nTTP (4.10s) (P < 0.001). The postoperative relative arterial component of the surgical hemisphere (median: 0.04) was significantly higher than the preoperative relative arterial component (median: - 0.02, P < 0.001). In contrast, the postoperative relative venous component of the surgical hemisphere (median: - 0.05) was significantly lower than the preoperative value (median: 0.05, P < 0.001). The improvement in cerebral perfusion parameters observed on postoperative DSC perfusion MRI demonstrated that stand-alone multiple burr hole surgery could be a favorable salvage revascularization technique after mEDAS failure in patients with ischemic MMD.

Research Note: Functional Connectivity Between a Corticostriatal Network and the Cerebellum.

Basal ganglia and cerebellum are structurally and functionally connected in animals. In humans, tractography and seed-based functional connectivity have confirmed this cerebellar-striatal relation. Independent component analysis (ICA) showed that both cerebellum and basal ganglia take part in distinct intrinsic networks. Probabilistic ICA analysis was applied to the brain images of 15 healthy volunteers during the resting state and using a 3T MRI. A spatial map corresponding to dorsal and ventral basal ganglia circuits was also found to be in functional coherence with crus 2, especially with its vermal region. It is speculated that such cerebellar-basal ganglionic rsFC could reflect structural interconnections traced in animals and explain reward-based activity detected in the cerebellum.

Radiation Pneumonitis in Thoracic Cancer Patients: Multi-Center Voxel-Based Analysis.

Simple SummaryThe pathophysiology of radiation pneumonitis (RP) after thoracic cancer radiation treatments is still not completely understood although the identification of underlying RP mechanisms may improve the therapeutic window of thoracic cancer patients. The aim of our retrospective study was to explore the dose–response patterns associated with RP by a multi-center voxel-based analysis. In a heterogeneously treated population of 382 thoracic cancer patients, we confirmed the previously described heart–lung interaction in the development of RP. The empowerment of VBA with a novel description of dose map spatial properties based on probabilistic independent component analysis (PICA) and connectograms provided valuable additional and independent information on the radiobiology of RP.This study investigates the dose–response patterns associated with radiation pneumonitis (RP) in patients treated for thoracic malignancies with different radiation modalities. To this end, voxel-based analysis (VBA) empowered by a novel strategy for the characterization of spatial properties of dose maps was applied. Data from 382 lung cancer and mediastinal lymphoma patients from three institutions treated with different radiation therapy (RT) techniques were analyzed. Each planning CT and biologically effective dose map (α/β = 3 Gy) was spatially normalized on a common anatomical reference. The VBA of local dose differences between patients with and without RP was performed and the clusters of voxels with dose differences that significantly correlated with RP at a p-level of 0.05 were generated accordingly. The robustness of VBA inference was evaluated by a novel characterization for spatial properties of dose maps based on probabilistic independent component analysis (PICA) and connectograms. This lays robust foundations to the obtained findings that the lower parts of the lungs and the heart play a prominent role in the development of RP. Connectograms showed that the dataset can support a radiobiological differentiation between the main heart and lung substructures.

Probing thoracic dose patterns associated to pericardial effusion and mortality in patients treated with photons and protons for locally advanced non-small-cell lung cancer

PurposeTo investigate thoracic dose–response patterns for pericardial effusion (PCE) and mortality in patients treated for locally advanced Non-Small-Cell Lung Cancer (NSCLC) by Intensity Modulated RT (IMRT) or Passive-Scattering Proton Therapy (PSPT). MethodsAmong 178 patients, 43.5% developed grade ≥ 2 PCE. Clinical and dosimetric factors associated with PCE or overall survival (OS) were identified via multi-variable Cox proportional hazards modeling. The Voxel-Based Analyses (VBAs) of local dose differences between patients with and without PCE and mortality was performed. The robustness of VBA results was assessed by a novel characterization of spatial properties of dose distributions based on probabilistic independent component analysis (PICA) and connectograms. ResultsSeveral non-dosimetric variables were selected by the multivariable analysis for the considered outcomes, while the time-dependent PCE onset was uncorrelated with the OS (p = 0.34) at a multi-variable Cox analysis. Despite the significant PSPT dosimetric advantage, the RT technique did not affect the occurrence of PCE or OS.VBAs highlighted largely overlapping clusters significantly associated with PCE endpoints in heart and lungs. No significant dosimetric patterns related to mortality endpoints were found.PICA identified 43 components homogeneously scattered within thorax, while connectograms showed modest correlations between doses in main cardio-pulmonary substructures. ConclusionsSpatially resolved analysis highlighted dose patterns related to radiation-induced cardiac toxiciy and the observed organ-based dose–response mismatch in PSPT and IMRT. Indeed, the thoracic regions spared by PSPT poorly overlapped with the areas involved in PCE development, as highlited by VBA. PICA and connectograms proved valuable tools for assessing the robusteness of obtained VBA inferences.

Increased Salience Network Connectivity Following Manual Therapy is Associated with Reduced Pain in Chronic Low Back Pain Patients

Chronic low back pain (cLBP) has been associated with changes in brain plasticity. Nonpharmacological therapies such as Manual Therapy (MT) have shown promise for relieving cLBP. However, translational neuroimaging research is needed to understand potential central mechanisms supporting MT. We investigated the effect of MT on resting-state salience network (SLN) connectivity, and whether this was associated with changes in clinical pain. Fifteen cLBP patients, and 16 matched healthy controls (HC) were scanned with resting functional Magnetic Resonance Imaging (fMRI), before and immediately after a MT intervention (cross-over design with two separate visits, pseudorandomized, grades V ‘Manipulation’ and III ‘Mobilization’ of the Maitland Joint Mobilization Grading Scale). Patients rated clinical pain (0–100) pre- and post-therapy. SLN connectivity was assessed using dual regression probabilistic independent component analysis. Both manipulation (Pre: 39.43 ± 16.5, Post: 28.43 ± 16.5) and mobilization (Pre: 38.83 ± 17.7, Post: 31.76 ± 19.4) reduced clinical back pain (P < .05). Manipulation (but not mobilization) significantly increased SLN connectivity to thalamus and primary motor cortex. Additionally, a voxelwise regression indicated that greater MT-induced increase in SLN connectivity to the lateral prefrontal cortex was associated with greater clinical back pain reduction immediately after intervention, for both manipulation (r = -0.8) and mobilization (r = -0.54). Our results suggest that MT is successful in reducing clinical low back pain by both spinal manipulation and spinal mobilization. Furthermore, this reduction post-manipulation occurs via modulation of SLN connectivity to sensorimotor, affective, and cognitive processing regions. PerspectiveMT both reduces clinical low back pain and modulates brain activity important for the processing of pain. This modulation was shown by increased functional brain connectivity between the salience network and brain regions involved in cognitive, affective, and sensorimotor processing of pain.

The laminar pattern of resting state in human cerebral cortex

Resting state functional Magnetic Resonance Imaging (RS-fMRI) provides the means to measure neuronal activity. One of the most commonly used methods to explore the RS-fMRI signal is the Probabilistic Independent Component Analysis (PICA). PICA allows to depict brain functional connectivity (FC) networks. Yet most of the IC maps obtained with this method do not represent any particular FC network. Consequently, those IC maps are classified as artifacts or noise of an unknown source. We hypothesized that the unexplained RS-fMRI signal patterns that are picked up by the PICA can be related to the differences in oxygen metabolism and blood flow in cortical layers. This study aimed at (1) providing preliminary evidence to the effects of laminar organization of neocortex on the RS-fMRI signal, and (2) evaluating the application of laminar maps to aid the classification of IC maps. We created laminar maps 1–4 that depict relative cortical thickness of layers IV and VI. Our data show that the RS-fMRI signal is significantly related to the relative thickness of the cortical layer VI but not layer IV. Importantly, the laminar maps 1–4 overlap with four separate IC maps. Thus, the laminar maps 1–4 improve classification and interpretation of the IC maps. Moreover, the laminar maps 1–4 may be considered as FC networks that are the bridging piece between particular cognitive functions. Together, these data provide preliminary evidence to the fundamental questions about the role of cortical layering in the RS-fMRI signal and brain FC networks.

Alterations in sympathetic and parasympathetic brain networks in obstructive sleep apnea

Objective/backgroundObstructive sleep apnea (OSA) patients experience hypoxia and, potentially, autonomic impairments stemming from neural damage. In this study, the executive control networks (ECNs), salience networks (SNs), and default mode networks (DMNs) of adult OSA patients, as well as their relationships with autonomic impairment, were investigated through independent component analysis (ICA). Patients/methodsA total of 41 OSA patients and 19 healthy controls volunteers were recruited and subjected to polysomnography to ascertain their degree, if any, of sleep apnea. Each participant also underwent a cardiovascular autonomic survey, with the participant's baroreflex sensitivity (BRS) being determined based on heart rate and blood pressure alterations. The resting fMRI data of the participants was separated using probabilistic ICA, and six autonomic resting-state networks were established for group comparisons. The differences in autonomic parameters, autonomic functional connectivity (FC), and clinical severity were then correlated. ResultsThe OSA group had significantly worse BRS values than the controls, as well as lower FC in the posterior and anterior SNs, bilateral ECNs, and the ventral DMN, and higher FC in the left ECN. These intrinsic connectivity networks showed dissociable correlations with greater baroreflex impairment and clinical disease severity. The higher FC in the left ECN was associated with the lower FC in the ventral DMN. ConclusionsOur findings suggest that autonomic dysfunction in OSA might be accompanied by central autonomic network alterations. The stronger sympathetic-associated regions in ECNs and the weaker parasympathetic-associated regions in DMNs may represent intrinsic neural architecture fluctuations underlining their consequent processes in OSA.

HINT: A hierarchical independent component analysis toolbox for investigating brain functional networks using neuroimaging data

BackgroundIndependent component analysis (ICA) is a popular tool for investigating brain organization in neuroscience research. In fMRI studies, an important goal is to study how brain networks are modulated by subjects’ clinical and demographic variables. Existing ICA methods and toolboxes don’t incorporate subjects’ covariates effects in ICA estimation of brain networks, which potentially leads to loss in accuracy and statistical power in detecting brain network differences between subjects’ groups. New methodWe introduce a Matlab toolbox, HINT (Hierarchical INdependent component analysis Toolbox), that provides a hierarchical covariate-adjusted ICA (hc-ICA) for modeling and testing covariate effects and generates model-based estimates of brain networks on both the population- and individual-level. HINT provides a user-friendly Matlab GUI that allows users to easily load images, specify covariate effects, monitor model estimation via an EM algorithm, specify hypothesis tests, and visualize results. HINT also has a command line interface which allows users to conveniently run and reproduce the analysis with a script. Comparison to existing methodsHINT implements a new multi-level probabilistic ICA model for group ICA. It provides a statistically principled ICA modeling framework for investigating covariate effects on brain networks. HINT can also generate and visualize model-based network estimates for user-specified subject groups, which greatly facilitates group comparisons. ResultsWe demonstrate the steps and functionality of HINT with an fMRI example data to estimate treatment effects on brain networks while controlling for other covariates. Results demonstrate estimated brain networks and model-based comparisons between the treatment and control groups. In comparisons using synthetic fMRI data, HINT shows desirable statistical power in detecting group differences in networks especially in small sample sizes, while maintaining a low false positive rate. HINT also demonstrates similar or increased accuracy in reconstructing both population- and individual-level source signal maps as compared to some state-of-the-art group ICA methods. ConclusionHINT can provide a useful tool for both statistical and neuroscience researchers to evaluate and test differences in brain networks between subject groups.

Non-Gaussian Penalized PARAFAC Analysis for fMRI Data

Independent Component Analysis (ICA) method has been used widely and successfully in functional magnetic resonance imaging (fMRI) data analysis for both single and group subjects. As an extension of the ICA, tensorial probabilistic ICA (TPICA) is used to decompose fMRI group data into three modes: subject, temporal and spatial. However, due to the independent constraint of the spatial components, TPICA is not very efficient in the presence of overlapping of active regions of different spatial components. Parallel factor analysis (PARAFAC) is another method used to process three-mode data and can be solved by alternating least-squares. PARAFAC may converge into some degenerate solutions if the matrix of one mode is collinear. It is reasonable to find significant collinear relationships within subject mode of two similar subjects in group fMRI data. Thus, both TPICA and PARAFAC have unavoidable drawbacks. In this paper, we try to alleviate both overlapping and collinearity issues by integrating the characters of PARAFAC and TPICA together by imposing a non-Gaussian penalty term to each spatial component under the PARAFAC framework. This proposed algorithm can then regulate the spatial components, as the high non-Gaussianity can possible avoid the degenerate solutions aroused by collinearity issue, and eliminate the independent constraint of the spatial components to bypass the overlapping issue. The algorithm outperforms TPICA and PARAFAC on the simulation data. The results of this algorithm on real fMRI data are also consistent with other algorithms.

Structural and Functional Reorganization of the Brain in Migraine Without Aura.

It remains unknown whether migraine headache has a progressive component in its pathophysiology. Quantitative MRI may provide valuable insight into abnormal changes in the migraine interictum and assist in identifying disrupted brain networks. We carried out a data-driven study of structural integrity and functional connectivity of the resting brain in migraine without aura. MRI scanning was performed in 36 patients suffering from episodic migraine without aura and 33 age-matched healthy subjects. Voxel-wise analysis of regional brain volume was performed by registration of the T1-weighted MRI scans into a common study brain template using the tensor-based morphometry (TBM) method. Changes in functional synchronicity of the brain networks were assessed using probabilistic independent component analysis (ICA). TBM revealed that migraine is associated with reduced volume of the medial prefrontal cortex (mPFC). Among 375 functional brain networks, resting-state connectivity was decreased between two components spanning the visual cortex, posterior insula, and parietal somatosensory cortex. Our study reveals structural and functional alterations of the brain in the migraine interictum that may stem from underlying disease risk factors and the “silent” aura phenomenon. Longitudinal studies will be needed to investigate whether interictal brain changes are progressive and associated with clinical disease trajectories.

Sampling Rate Effects on Resting State fMRI Metrics.

Low image sampling rates used in resting state functional magnetic resonance imaging (rs-fMRI) may cause aliasing of the cardiorespiratory pulsations over the very low frequency (VLF) BOLD signal fluctuations which reflects to functional connectivity (FC). In this study, we examine the effect of sampling rate on currently used rs-fMRI FC metrics. Ultra-fast fMRI magnetic resonance encephalography (MREG) data, sampled with TR 0.1 s, was downsampled to different subsampled repetition times (sTR, range 0.3–3 s) for comparisons. Echo planar k-space sampling (TR 2.15 s) and interleaved slice collection schemes were also compared against the 3D single shot trajectory at 2.2 s sTR. The quantified connectivity metrics included stationary spatial, time, and frequency domains, as well as dynamic analyses. Time domain methods included analyses of seed-based functional connectivity, regional homogeneity (ReHo), coefficient of variation, and spatial domain group level probabilistic independent component analysis (ICA). In frequency domain analyses, we examined fractional and amplitude of low frequency fluctuations. Aliasing effects were spatially and spectrally analyzed by comparing VLF (0.01–0.1 Hz), respiratory (0.12–0.35 Hz) and cardiac power (0.9–1.3 Hz) FFT maps at different sTRs. Quasi-periodic pattern (QPP) of VLF events were analyzed for effects on dynamic FC methods. The results in conventional time and spatial domain analyses remained virtually unchanged by the different sampling rates. In frequency domain, the aliasing occurred mainly in higher sTR (1–2 s) where cardiac power aliases over respiratory power. The VLF power maps suffered minimally from increasing sTRs. Interleaved data reconstruction induced lower ReHo compared to 3D sampling (p < 0.001). Gradient recalled echo-planar imaging (EPI BOLD) data produced both better and worse metrics. In QPP analyses, the repeatability of the VLF pulse detection becomes linearly reduced with increasing sTR. In conclusion, the conventional resting state metrics (e.g., FC, ICA) were not markedly affected by different TRs (0.1–3 s). However, cardiorespiratory signals showed strongest aliasing in central brain regions in sTR 1–2 s. Pulsatile QPP and other dynamic analyses benefit linearly from short TR scanning.

(270) Increased Salience Network Connectivity following Spinal Manipulative Therapy is Associated with Reduced Pain in Chronic Low Back Pain Patients

Chronic low back pain (cLBP) has been associated with maladaptive brain plasticity, and non-pharmacological therapies such as Spinal Manipulative Therapy (SMT) have shown promise in reducing pain and modifying intrinsic brain physiology. However, more translational research is needed using clinical models. We investigated functional salience network (SLN) connectivity response to SMT, and linked response to cLBP pain reduction. Fifteen cLBP patients (8 female, 37.7 ± 9.7 (M± SD) years old), and 16 healthy controls (HC, 8 female, 38.2 ± 10.4 years old) were scanned with resting state fMRI before and after a single session of spinal manipulation and spinal mobilization (grades V and III of the Maitland Joint Mobilization Grading Scale) at separate visits. Patients rated clinical pain (0-100) pre- and post-therapy. Intrinsic SLN connectivity was assessed using dual regression probabilistic independent component analysis. Voxelwise analyses were cluster corrected for multiple comparison (z>2.3, p

Probabilistic independent component analysis of dynamic susceptibility contrast perfusion MRI in metastatic brain tumors

PurposeTo identify clinically relevant magnetic resonance imaging (MRI) features of different types of metastatic brain lesions, including standard anatomical, diffusion weighted imaging (DWI) and dynamic susceptibility contrast (DSC) perfusion MRI.MethodsMRI imaging was retrospectively assessed on one hundred and fourteen (N = 114) brain metastases including breast (n = 27), non-small cell lung cancer (NSCLC, n = 43) and ‘other’ primary tumors (n = 44). Based on 114 patient’s MRI scans, a total of 346 individual contrast enhancing tumors were manually segmented. In addition to tumor volume, apparent diffusion coefficients (ADC) and relative cerebral blood volume (rCBV) measurements, an independent component analysis (ICA) was performed with raw DSC data in order to assess arterio-venous components and the volume of overlap (AVOL) relative to tumor volume, as well as time to peak (TTP) of T2* signal from each component.ResultsResults suggests non-breast or non-NSCLC (‘other’) tumors had higher volume compare to breast and NSCLC patients (p = 0.0056 and p = 0.0003, respectively). No differences in median ADC or rCBV were observed across tumor types; however, breast and NSCLC tumors had a significantly higher “arterial” proportion of the tumor volume as indicated by ICA (p = 0.0062 and p = 0.0018, respectively), while a higher “venous” proportion were prominent in breast tumors compared with NSCLC (p = 0.0027) and ‘other’ lesions (p = 0.0011). The AVOL component was positively related to rCBV in all groups, but no correlation was found for arterial and venous components with respect to rCBV values. Median time to peak of arterial and venous components were 8.4 s and 12.6 s, respectively (p < 0.0001). No difference was found in arterial or venous TTP across groups.ConclusionsAdvanced ICA-derived component analysis demonstrates perfusion differences between metastatic brain tumor types that were not observable with classical ADC and rCBV measurements. These results highlight the complex relationship between brain tumor vasculature characteristics and the site of primary tumor diagnosis.

Machine learning identifies "rsfMRI epilepsy networks" in temporal lobe epilepsy.

Experimental models have provided compelling evidence for the existence of neural networks in temporal lobe epilepsy (TLE). To identify and validate the possible existence of resting-state "epilepsy networks," we used machine learning methods on resting-state functional magnetic resonance imaging (rsfMRI) data from 42 individuals with TLE. Probabilistic independent component analysis (PICA) was applied to rsfMRI data from 132 subjects (42 TLE patients + 90 healthy controls) and 88 independent components (ICs) were obtained following standard procedures. Elastic net-selected features were used as inputs to support vector machine (SVM). The strengths of the top 10 networks were correlated with clinical features to obtain "rsfMRI epilepsy networks." SVM could classify individuals with epilepsy with 97.5% accuracy (sensitivity = 100%, specificity = 94.4%). Ten networks with the highest ranking were found in the frontal, perisylvian, cingulo-insular, posterior-quadrant, thalamic, cerebello-thalamic, and temporo-thalamic regions. The posterior-quadrant, cerebello-thalamic, thalamic, medial-visual, and perisylvian networks revealed significant correlation (r> 0.40) with age at onset of seizures, the frequency of seizures, duration of illness, and a number of anti-epileptic drugs. IC-derived rsfMRI networks contain epilepsy-related networks and machine learning methods are useful in identifying these networks in vivo. Increased network strength with disease progression in these "rsfMRI epilepsy networks" could reflect epileptogenesis in TLE. • ICA of resting-state fMRI carries disease-specific information about epilepsy. • Machine learning can classify these components with 97.5% accuracy. • "Subject-specific epilepsy networks" could quantify "epileptogenesis" in vivo.

A study of neural activity and functional connectivity within the olfactory brain network in Parkinson's disease

Olfactory dysfunction is an early manifestation of Parkinson's disease (PD). The present study aimed to illustrate potential differences between PD patients and healthy controls in terms of neural activity and functional connectivity within the olfactory brain network. Twenty PD patients and twenty healthy controls were examined with olfactory fMRI and resting-state fMRI. Data analysis of olfactory fMRI included data-driven tensorial independent component (ICA) and task-driven general linear model (GLM) analyses. Data analysis of resting-state fMRI included probabilistic ICA based on temporal concatenation and functional connectivity analysis within the olfactory network. ICA of olfactory fMRI identified an olfactory network consisting of the posterior piriform cortex, insula, right orbitofrontal cortex and thalamus. Recruitment of this network was less significant for PD patients. GLM analysis revealed significantly lower activity in the insula bilaterally and the right orbitofrontal cortex in PD compared to healthy controls but no significant differences in the olfactory cortex itself. Analysis of resting-state fMRI did not reveal any differences in the functional connectivity within the olfactory, default mode, salience or central executive networks between the two groups. In conclusion, olfactory dysfunction in PD is associated with less significant recruitment of the olfactory brain network. ICA could demonstrate differences in both the olfactory cortex and its main projections, compared to GLM that revealed differences only on the latter. Resting-state fMRI did not reveal any significant differences in functional connectivity within the olfactory, default mode, salience and central executive networks in this cohort.

Random forest based classification of alcohol dependence patients and healthy controls using resting state MRI

Currently, classification of alcohol use disorder (AUD) is made on clinical grounds; however, robust evidence shows that chronic alcohol use leads to neurochemical and neurocircuitry adaptations. Identifications of the neuronal networks that are affected by alcohol would provide a more systematic way of diagnosis and provide novel insights into the pathophysiology of AUD. In this study, we identified network-level brain features of AUD, and further quantified resting-state within-network, and between-network connectivity features in a multivariate fashion that are classifying AUD, thus providing additional information about how each network contributes to alcoholism.Resting-state fMRI were collected from 92 individuals (46 controls and 46 AUDs). Probabilistic Independent Component Analysis (PICA) was used to extract brain functional networks and their corresponding time-course for AUD and controls. Both within-network connectivity for each network and between-network connectivity for each pair of networks were used as features. Random forest was applied for pattern classification.The results showed that within-networks features were able to identify AUD and control with 87.0% accuracy and 90.5% precision, respectively. Networks that were most informative included Executive Control Networks (ECN), and Reward Network (RN). The between-network features achieved 67.4% accuracy and 70.0% precision. The between-network connectivity between RN-Default Mode Network (DMN) and RN-ECN contribute the most to the prediction.In conclusion, within-network functional connectivity offered maximal information for AUD classification, when compared with between-network connectivity. Further, our results suggest that connectivity within the ECN and RN are informative in classifying AUD. Our findings suggest that machine-learning algorithms provide an alternative technique to quantify large-scale network differences and offer new insights into the identification of potential biomarkers for the clinical diagnosis of AUD.