Machine learning approach for obstructive sleep apnea screening using brain diffusion tensor imaging.

Patients with obstructive sleep apnea (OSA) show autonomic, mood, cognitive, and breathing dysfunctions that are linked to increased morbidity and mortality, which can be improved with early screening and intervention. The gold standard and other available methods for OSA diagnosis are complex, require whole-night data, and have significant wait periods that potentially delay intervention. Our aim was to examine whether using faster and less complicated machine learning models, including support vector machine (SVM) and random forest (RF), with brain diffusion tensor imaging (DTI) data can classify OSA from healthy controls. We collected two DTI series from 59 patients with OSA [age: 50.2 ± 9.9 years; body mass index (BMI): 31.5 ± 5.6 kg/m2 ; apnea-hypopnea index (AHI): 34.1 ± 21.2 events/h 23 female] and 96 controls (age: 51.8 ± 9.7 years; BMI: 26.2 ± 4.1 kg/m2 ; 51 female) using a 3.0-T magnetic resonance imaging scanner. Using DTI data, mean diffusivity maps were calculated from each series, realigned and averaged, normalised to a common space, and used to conduct cross-validation for model training and selection and to predict OSA. The RF model showed 0.73 OSA and controls classification accuracy and 0.85 area under the curve (AUC) value on the receiver-operator curve. Cross-validation showed the RF model with comparable fitting over SVM for OSA and control data (SVM; accuracy, 0.77; AUC, 0.84). The RF ML model performs similar to SVM, indicating the comparable statistical fitness to DTI data. The findings indicate that RF model has similar AUC and accuracy over SVM, and either model can be used as a faster OSA screening tool for subjects having brain DTI data.

Non-Gaussian Diffusion Imaging Shows Brain Myelin and Axonal Changes in Obstructive Sleep Apnea.

OBJECTIVE: Obstructive sleep apnea (OSA) is accompanied by brain changes in areas that regulate autonomic, cognitive, and mood functions, which were initially examined by Gaussian-based diffusion tensor imaging measures, but can be better assessed with non-Gaussian measures. We aimed to evaluate axonal and myelin changes in OSA using axial (AK) and radial kurtosis (RK) measures. MATERIALS AND METHODS: We acquired diffusion kurtosis imaging data from 22 OSA and 26 controls; AK and RK maps were calculated, normalized, smoothed, and compared between groups using analysis of covariance. RESULTS: Increased AK, indicating axonal changes, emerged in the insula, hippocampus, amygdala, dorsolateral pons, and cerebellar peduncles and showed more axonal injury over previously identified damage. Higher RK, showing myelin changes, appeared in the hippocampus, amygdala, temporal and frontal lobes, insula, midline pons, and cerebellar peduncles and showed more widespread myelin damage over previously identified injury. CONCLUSIONS: Axial kurtosis and RK measures showed widespread changes over Gaussian-based techniques, suggesting a more sensitive nature of kurtoses to injury.

Associations between brain white matter integrity and disease severity in obstructive sleep apnea.

Obstructive sleep apnea (OSA) is characterized by recurrent upper airway blockage, with continued diaphragmatic efforts to breathe during sleep. Brain structural changes in OSA appear in various regions, including white matter sites that mediate autonomic, mood, cognitive, and respiratory control. However, the relationships between brain white matter changes and disease severity in OSA are unclear. This study examines associations between an index of tissue integrity, magnetization transfer (MT) ratio values (which show MT between free and proton pools associated with tissue membranes and macromolecules), and disease severity (apnea-hypopnea index [AHI]) in OSA subjects. We collected whole-brain MT imaging data from 19 newly diagnosed, treatment-naïve OSA subjects (50.4 ± 8.6 years of age, 13 males, AHI 39.7 ± 24.3 events/hr], using a 3.0-Tesla MRI scanner. With these data, whole-brain MT ratio maps were calculated, normalized to common space, smoothed, and correlated with AHI scores by using partial correlation analyses (covariates, age and gender; P < 0.005). Multiple brain sites in OSA subjects, including superior and inferior frontal regions, ventral medial prefrontal cortex and nearby white matter, midfrontal white matter, insula, cingulate and cingulum bundle, internal and external capsules, caudate nuclei and putamen, basal forebrain, hypothalamus, corpus callosum, and temporal regions, showed principally lateralized negative correlations (P < 0.005). These regions showed significant correlations even with correction for multiple comparisons (cluster-level, family-wise error, P < 0.05), except for a few superior frontal areas. Predominantly negative correlations emerged between local MT values and OSA disease severity, indicating potential usefulness of MT imaging for examining the OSA condition. These findings indicate that OSA severity plays a significant role in white matter injury. © 2016 Wiley Periodicals, Inc.

Impaired Glymphatic System Actions in Obstructive Sleep Apnea Adults.

Study Objectives: Obstructive sleep apnea (OSA) is accompanied by sleep fragmentation and altered sleep architecture, which can potentially hinder the glymphatic system, increasing risks for Alzheimer's disease (AD), but the status is unclear in OSA. Our aim was to investigate the glymphatic system in OSA subjects and examine the relationships between OSA disease severity, sleep symptoms, and glymphatic system indices in OSA using diffusion tensor imaging (DTI). Methods: We acquired DTI data from 59 OSA and 62 controls using a 3.0-Tesla MRI and examined OSA disease severity and sleep symptoms with the Pittsburgh Sleep Quality Index (PSQI) and Epworth Sleepiness Scale (ESS). Diffusivity maps in the x-axis (Dxx), y-axis (Dyy), and z-axis (Dzz), as well as in x-y axis (Dxy), y-z axis (Dyz), and x-z axis (Dxz) were calculated, diffusion values for the projection and association fibers extracted, and the DTI analyses along the perivascular space (DTI-ALPS index) were performed. The glymphatic system indices were compared between groups and correlated with disease severity and sleep symptoms in OSA subjects. Results: Dzz values, derived from projection fiber areas, Dyy and Dzz values from association fiber areas, as well as ALPS and Dyzmean values were significantly reduced in OSA over controls. Significant correlations emerged between disease severity, sleep symptoms, and Dxy, Dxx, and Dzz values in OSA subjects. Conclusion: OSA patients show abnormal glymphatic system function that may contribute to increased risks for AD. The findings suggest that the APLS method can be used to assess the glymphatic system in OSA patients.

Brain regional homogeneity changes after short-term positive airway pressure treatment in patients with obstructive sleep apnea.

Patients with obstructive sleep apnea (OSA) reveal functional changes in brain sites involved in autonomic, cognitive, and mood regulations. However, it is unclear whether these brain changes reverse with short-term positive airway pressure (PAP) treatment. Our aim was to examine brain functional changes in response to 3-months of PAP treatment using regional homogeneity (ReHo) measures, where increased and decreased ReHo value indicates hyper- and hypo-local neural activities, respectively, and considered as functional deficits. We collected brain magnetic resonance imaging data as well as mood, cognitive, and sleep variables from 17 treatment-naïve OSA at baseline and after 3-months of PAP treatment and 25 age- and gender-matched healthy controls. Whole-brain ReHo maps were calculated and compared between OSA and controls and OSA subjects before and after PAP treatment. At baseline, treatment-naïve OSA subjects showed higher ReHo in the bilateral thalamus, putamen, postcentral gyrus, paracentral lobule, supplementary motor area, and right insula, and lower ReHo in the frontal and parietal cortices, compared to controls. After 3-months of PAP treatment, abnormal sleep and mood scores decreased significantly to normal levels. ReHo decreased in the autonomic and somatosensory control areas, including the thalamus, putamen, postcentral gyrus, and insula, and increased in the cognitive and affective regulatory parietal regions. The normalized ReHo was correlated with improved sleep quality and reduced anxiety symptoms. These findings suggest that 3-months of PAP use can improve sleep, mood issues, and partly recover brain activities, however, longer PAP treatment may be required to fully and permanently reverse brain functional deficits.

Altered resting-state hippocampal and caudate functional networks in patients with obstructive sleep apnea.

INTRODUCTION: Brain structural injury and metabolic deficits in the hippocampus and caudate nuclei may contribute to cognitive and emotional deficits found in obstructive sleep apnea (OSA) patients. If such contributions exist, resting-state interactions of these subcortical sites with cortical areas mediating affective symptoms and cognition should be disturbed. Our aim was to examine resting-state functional connectivity (FC) of the hippocampus and caudate to other brain areas in OSA relative to control subjects, and to relate these changes to mood and neuropsychological scores. METHODS: We acquired resting-state functional magnetic resonance imaging (fMRI) data from 70 OSA and 89 healthy controls using a 3.0-Tesla magnetic resonance imaging scanner, and assessed psychological and behavioral functions, as well as sleep issues. After standard fMRI data preprocessing, FC maps were generated for bilateral hippocampi and caudate nuclei, and compared between groups (ANCOVA; covariates, age and gender). RESULTS: Obstructive sleep apnea subjects showed significantly higher levels of anxiety and depressive symptoms over healthy controls. In OSA subjects, the hippocampus showed disrupted FC with the thalamus, para-hippocampal gyrus, medial and superior temporal gyrus, insula, and posterior cingulate cortex. Left and right caudate nuclei showed impaired FC with the bilateral inferior frontal gyrus and right angular gyrus. In addition, altered limbic-striatal-cortical FC in OSA showed relationships with behavioral and neuropsychological variables. CONCLUSIONS: The compromised hippocampal-cortical FC in OSA may underlie depression and anxious mood levels in OSA, while impaired caudate-cortical FC may indicate deficits in reward processing and cognition. These findings provide insights into the neural mechanisms underlying the comorbidity of mood and cognitive deficits in OSA.

Correlations between Waist and Neck Circumferences and Obstructive Sleep Apnea Characteristics.

PURPOSE: The body mass index (BMI), an estimate of body fat, provides a rather imprecise indication of risk for obstructive sleep apnea (OSA). We examined whether other measures, including waist and neck circumference, provide improved indicators of risk in treatment-naïve OSA subjects. METHODS: We studied 59 OSA subjects [age, 48.8±10.0 years; BMI, 31.9±6.6 kg/m2; apnea-hypopnea-index (AHI), 38.5±23.0 events/hour; sleep efficiency index (SEI, n=52), 78.6±14.4%; lowest oxygen saturation (SaO2 nadir), 79.5±8.0%; systolic blood pressure (BP), 127.4±15.7 mmHg; diastolic BP, 80.1±9.1 mmHg; 43 male), and determined waist and neck circumferences (waist, 107.4±15.3 cm; neck, 41.8±4.7 cm), daytime sleepiness [Epworth sleepiness scale (ESS), 8.7±4.6], sleep quality [Pittsburgh sleep quality index (PSQI), 8.5±4.1], depression levels [Beck depression inventory II (BDI-II), 6.6±5.7), and anxiety levels [Beck anxiety inventory (BAI), 6.2±7.2]. We used partial correlation procedures (covariates, age and gender) to examine associations between BMI, waist, and neck circumferences vs. AHI, sleep, and neuropsychological variables. RESULTS: BMI, waist, and neck circumferences were significantly correlated with SaO2 nadir (BMI; r=-0.423, p=0.001; waist; r=-0.457, p<0.001; neck; r=-0.263, p=0.048), AHI (BMI; r=0.349, p=0.008; waist; r=0.459, p<0.001; neck; r=0.276, p=0.038), and systolic BP (BMI; r=0.354, p=0.007; waist; r=0.321, p=0.015; neck; r=0.388, p=0.003). SEI was significantly correlated with waist circumference (r=0.28, p=0.049), but higher with BMI (r=0.291, p=0.04). CONCLUSIONS: No other significant waist or neck correlations emerged. This study suggests that waist and neck measures correlate better than BMI with select disease severity (SaO2 nadir and AHI) in OSA subjects. The findings offer an easily-measured, ancillary means to assess OSA risk.

Global and Regional Brain Non-Gaussian Diffusion Changes in Newly Diagnosed Patients with Obstructive Sleep Apnea.

STUDY OBJECTIVES: Obstructive sleep apnea (OSA) patients show brain structural injury and functional deficits in autonomic, affective, and cognitive regulatory sites, as revealed by mean diffusivity (MD) and other imaging procedures. The time course and nature of gray and white matter injury can be revealed in more detail with mean kurtosis (MK) procedures, which can differentiate acute from chronic injury, and better show extent of damage over MD procedures. Our objective was to examine global and regional MK changes in newly diagnosed OSA, relative to control subjects. METHODS: Two diffusion kurtosis image series were collected from 22 recently-diagnosed, treatment-naïve OSA and 26 control subjects using a 3.0-Tesla MRI scanner. MK maps were generated, normalized to a common space, smoothed, and compared voxel-by-voxel between groups using analysis of covariance (covariates; age, sex). RESULTS: No age or sex differences appeared, but body mass index, sleep, neuropsychologic, and cognitive scores significantly differed between groups. MK values were significantly increased globally in OSA over controls, and in multiple localized sites, including the basal forebrain, extending to the hypothalamus, hippocampus, thalamus, insular cortices, basal ganglia, limbic regions, cerebellar areas, parietal cortices, ventral temporal lobe, ventrolateral medulla, and midline pons. Multiple sites, including the insular cortices, ventrolateral medulla, and midline pons showed more injury over previously identified damage with MD procedures, with damage often lateralized. CONCLUSIONS: Global mean kurtosis values are significantly increased in obstructive sleep apnea (OSA), suggesting acute tissue injury, and these changes are principally localized in critical sites mediating deficient functions in the condition. The mechanisms for injury likely include altered perfusion and hypoxemia-induced processes, leading to acute tissue changes in recently diagnosed OSA.

Disrupted functional brain network organization in patients with obstructive sleep apnea.

INTRODUCTION: Obstructive sleep apnea (OSA) subjects show impaired autonomic, affective, executive, sensorimotor, and cognitive functions. Brain injury in OSA subjects appears in multiple sites regulating these functions, but the integrity of functional networks within the regulatory sites remains unclear. Our aim was to examine the functional interactions and the complex network organization of these interactions across the whole brain in OSA, using regional functional connectivity (FC) and brain network topological properties. METHODS: We collected resting-state functional magnetic resonance imaging (MRI) data, using a 3.0-Tesla MRI scanner, from 69 newly diagnosed, treatment-naïve, moderate-to-severe OSA (age, 48.3 ± 9.2 years; body mass index, 31 ± 6.2 kg/m(2); apnea-hypopnea index (AHI), 35.6 ± 23.3 events/h) and 82 control subjects (47.6 ± 9.1 years; body mass index, 25.1 ± 3.5 kg/m(2)). Data were analyzed to examine FC in OSA over controls as interregional correlations and brain network topological properties. RESULTS: Obstructive sleep apnea subjects showed significantly altered FC in the cerebellar, frontal, parietal, temporal, occipital, limbic, and basal ganglia regions (FDR, P < 0.05). Entire functional brain networks in OSA subjects showed significantly less efficient integration, and their regional topological properties of functional integration and specialization characteristics also showed declined trends in areas showing altered FC, an outcome which would interfere with brain network organization (P < 0.05; 10,000 permutations). Brain sites with abnormal topological properties in OSA showed significant relationships with AHI scores. CONCLUSIONS: Our findings suggest that the dysfunction extends to resting conditions, and the altered FC and impaired network organization may underlie the impaired responses in autonomic, cognitive, and sensorimotor functions. The outcomes likely result from the prominent structural changes in both axons and nuclear structures, which occur in the condition.

Aberrant Insular Functional Network Integrity in Patients with Obstructive Sleep Apnea.

STUDY OBJECTIVES: Obstructive sleep apnea (OSA) is accompanied by tissue injury to the insular cortices, areas that regulate autonomic pain, dyspnea, and mood, all of which are affected in the syndrome. Presumably, the dysregulation of insular-related functions are mediated by aberrant functional connections with other brain regions; however, the integrity of the functional connectivity (FC) to other sites is undescribed. Our aim was to examine resting-state FC of the insular cortices to other brain areas in OSA, relative to control subjects. METHODS: We collected resting-state functional magnetic resonance imaging (MRI) data from 67 newly diagnosed, treatment-naïve OSA and 75 control subjects using a 3.0-Tesla MRI scanner. After standard processing, data were analyzed for the left and right insular FC. RESULTS: OSA subjects showed complex aberrant insular FC to several brain regions, including frontal, parietal, cingulate, temporal, limbic, basal ganglia, thalamus, occipital, cerebellar, and brainstem regions. Areas of altered FC in OSA showed linear relationships with magnitudes of sleep related and neuropsychologic-related variables, whereas control subjects showed no such relationships with those measures. CONCLUSIONS: Brain functional connections from insular sites to other brain regions in OSA subjects represent abnormal autonomic, affective, sensorimotor, and cognitive control networks that may affect both impaired parasympathetic and sympathetic interactions, as well as abnormal sensorimotor integration, affected in the condition. The functional changes likely result from the previously reported structural changes in OSA subjects, as demonstrated by diverse neuroimaging studies.

Anomalous inferior Vena cava as the cause of multiple deep venous thrombosis.

An anomalous Inferior Vena Cava (IVC) is a possible independent risk factor for deep vein thrombosis (DVT). This case represents the rare complication of an anomalous IVC causing multiple DVTs, not only in the lower extremity, but also in the abdominal periaortic circulation. In young patients who develop a DVT without risk factors, an anomalous IVC should be in the differential diagnosis.

Water Exchange across the Blood-Brain Barrier in Obstructive Sleep Apnea: An MRI Diffusion-Weighted Pseudo-Continuous Arterial Spin Labeling Study.

BACKGROUND AND PURPOSE: Obstructive sleep apnea (OSA) subjects show brain injury in sites that control autonomic, cognitive, and mood functions that are deficient in the condition. The processes contributing to injury may include altered blood-brain barrier (BBB) actions. Our aim was to examine BBB function, based on diffusion-weighted pseudo-continuous arterial spin labeling (DW-pCASL) procedures, in OSA compared to controls. METHODS: We performed DW-pCASL imaging in nine OSA and nine controls on a 3.0-Tesla MRI scanner. Global mean gray and white matter arterial transient time (ATT, an index of large artery integrity), water exchange rate across the BBB (Kw, BBB function), DW-pCASL ratio, and cerebral blood flow (CBF) values were compared between OSA and control subjects. RESULTS: Global mean gray and white matter ATT (OSA vs. controls; gray matter, 1.691 ± .120 vs. 1.658 ± .109 second, P = .49; white matter, 1.700 ± .115 vs. 1.650 ± .114 second, P = .44), and CBF values (gray matter, 57.4 ± 15.8 vs. 58.2 ± 10.7 ml/100 g/min, P = .67; white matter, 24.2 ± 7.0 vs. 24.6 ± 6.7 ml/100 g/min, P = .91) did not differ significantly, but global gray and white matter Kw (gray matter, 158.0 ± 28.9 vs. 220.8 ± 40.6 min(-1) , P = .002; white matter, 177.5 ± 57.2 vs. 261.1 ± 51.0 min(-1) , P = .006), and DW-pCASL ratio (gray matter, .727 ± .076 vs. .823 ± .069, P = .011; white matter, .722 ± .144 vs. .888 ± .100, P = .004) values were significantly reduced in OSA over controls. CONCLUSIONS: OSA subjects show compromised BBB function, but intact large artery integrity. The BBB alterations may introduce neural damage contributing to abnormal functions in OSA, and suggest a need to repair BBB function with strategies commonly used in other fields.