Using principal component analysis to determine which vestibular stimuli provide best biomarkers for separating Alzheimer's from mixed Alzheimer's disease.

Alzheimer's disease (AD) is often mixed with cerebrovascular disease (AD-CVD). Heterogeneity of dementia etiology and the overlapping of neuropathological features of AD and AD-CVD make feature identification of the two challenging. Separation of AD from AD-CVD is important as the optimized treatment for each group may differ. Recent studies using vestibular responses recorded from electrovestibulography (EVestG™) have offered promising results for separating these two pathologies. An EVestG measurement records responses to several different physical stimuli (called tilts). In previous research, the number of EVestG features from different tilts was selected based on physiological intuition to classify AD from AD-CVD. As the number of potential characteristic features from all tilts can be very large, in this study, we used an algorithm based on principal component analysis (PCA) to rank the most effective vestibular stimuli for differentiating AD from AD-CVD. Analyses were performed on the EVestG signals of 28 individuals with AD and 24 with AD-CVD. The results of this study showed that tilts simulating the otolithic organs (utricle and saccule) generated the most characteristic features for separating AD from AD-CVD.

Application of Vestibular Spontaneous Response as a Diagnostic Aid for Meniere's Disease.

In this paper, we report on a new method for assisting in Meniere's disease diagnosis. An accurate diagnosis of Meniere's is challenging, and requires an expert opinion after observing several clinical assessments and tests over a period of time. Our proposed method is based on the analysis of the spontaneous and driven ear evoked responses recorded using Electrovestibulography (EVestG). We used the EVestG signals of 35 individuals suspected of Meniere's and 26 age-matched healthy controls, out of which data of 14 patients with Meniere's and 16 healthy controls were used for developing the diagnostic algorithm (training set) and the rest for testing. While recording and analyzing the test dataset, the researchers were only aware the patients suffered some dizziness, and were kept blind to the exact diagnoses till the end of study. EVestG field potentials (FPs) and their firing pattern, in response to several whole body tilt stimuli from both left and right ears were extracted. We investigated several features of the extracted FPs in response to each of side, back/forward, rotation, up/down, supine rotation, and supine up/down tilt stimulations, and selected the top five features showing the most significant differences between of the groups of the training set for every tilt. An ad-hoc average voting classifier was designed based on building five single-feature classifiers (using Linear Discriminant analysis) and taking the average of the single-feature classifiers' votes. The results showed the side tilt data were best for the purpose of Meniere's diagnosis; it resulted in 78% and 90% sensitivity and specificity for test dataset, respectively. The second best accuracy was achieved using back/forward tilt. The results and their implications are discussed. Overall, the EVestG side tilt results encourage the use of vestibular response as a non-invasive, robust and quick screening for Meniere's and separating it from other types of dizziness.

Upper airway resistance changes from inspiration to expiration during wakefulness is a predictor of sleep apnea: A pilot study.

Upper airway structure is known to be affected in individuals with obstructive sleep apnea (OSA). In this pilot study, we investigated whether the changes of upper airway resistance (UAR) within a breathing phase, estimated from tracheal breathing sounds and airflow, is a predictor of OSA severity. Tracheal breathing sounds and airflow of 15 individuals in three groups of mild, moderate and severe OSA during wakefulness, recorded in supine position, were used in this study. The grouping of the study subjects were done based on their apnea/hypopnea index (AHI) per hour, prospectively. The breathing sounds and airflow were recorded simultaneously with full overnight polysomnography (PSG) assessment. The sounds of 3 noise-free breathing cycles were extracted and sequestered into inspiratory and expiratory phase segments manually for each study subject. The power spectra of the sound signals of each respiratory phase were calculated in 15ms windows with 50% overlap between adjacent windows. The UAR was then estimated as the ratio of the energy of power spectrum of each window within 100 to 2500 Hz, and its corresponding airflow. Then, the variance of the short-time windows' UAR, Var(UAR) during each respiratory phase was calculated. The ratio of the Var(UAR) in logarithmic scale between inspiration and expiration was found to be significantly different (p<;0.01) between the three OSA groups; also it was found to be highly correlated with AHI. These results are congruent with the hypothesis that the upper airway patency is not kept constant in OSA population, and suggest the change in acoustic UAR during wakefulness maybe considered as a predictor of OSA.

Investigating the feasibility of EVestG assessment for screening concussion.

Electrovestibulography (EVestG™) is a new technology that objectively measures the vestibular response. It has the potential to objectively, quickly and cost-effectively screen concussion. EVestG signals are recorded painlessly and non-invasively from the external ear in response to vestibular stimuli, and consist of brainstem and peripheral sensory oto-acoustic signals modulated by the cortical responses. In this study, we investigated the relationship between characteristic features of the extracted field potentials (FPs) of EVestG signals in people with side-impact concussion in comparison with those of control participants. 10 side-impact concussed individuals (4 Right and 6 left side-impact) and 10 age-and-gender-matched controls were tested by EVestG. The participants also completed comprehensive neuropsychological assessments. Characteristic features were extracted from the FPs during side tilt, and linear discriminant analysis (LDA) classification was applied to the extracted features using a leave-one-out routine. The results show the difference between the left and right FP area was significantly (P<0.05) different. The LDA classification resulted a sensitivity of 85% and specificity of 69% for separating concussed individuals from controls. EVestG appears to have diagnostic potential in diagnosing side impact concussion.

Tendency to overestimate the explicit time interval in relation to aging and cognitive decline.

Age-related deficits in explicit time perception has been reported by some studies. However, the findings are inconsistent about the preference of older adults to over/underestimate the observed interval as well as the relationship between the time estimation and the participant's cognitive status. In this study, we used a verbal estimation task for the rotation time of a virtual building (40 seconds) to assess the explicit interval timing of participants. The performance of a cohort of 250 cognitively-healthy adults and 10 Alzheimer's patients was analyzed in relation to their age and cognitive scale, measured by Montreal Cognitive Assessment (MoCA) score. The participants' performances were evaluated based on three measurements: Coefficient of variation (CV) for measuring stability, Absolute Error (AE) for measuring accuracy and Directional Error (DE) for measuring the degree of over/under-estimation. A significant difference was observed between the participants who overestimated the interval and those who underestimated it in terms of age, cognitive status and Absolute Error. We also found a significant effect of time estimation, with underestimation by cognitively healthy participants to mild over-estimation by 70+ year old and low-MoCA (MoCA score <; 26) participants as well as severe overestimation by Alzheimer's disease patients. The result of regression analysis for predicting MoCA score based on the dependent variables (AE, DE and CV) support the superiority of Directional Error to Absolute Error and Coefficient of Variation that are commonly used in the time perception studies.

Breathing sounds characteristics correlate with structural changes of upper airway due to obstructive sleep apnea.

This paper reports on a pilot study of acoustic properties of respiratory sounds in relation to obstructive sleep apnea (OSA). Data were breathing sounds of 60 individuals in three groups of non-, Moderate and Severe OSA using the apnea/hypopnea index (AHI) (20 in each group). The sounds power spectral energy and phase responses were calculated for each respiratory phase, and investigated among the groups. The results of power spectral analysis, congruent with imaging studies' findings, suggest a regional narrowing in upper airway that shows its effect in low and high frequencies as an opposite pattern between the OSA and non-OSA groups. The results of phase analysis also support an increased nonlinearity in high frequencies that is speculated to be related to the regional narrowing and thus regional elasticity of the airway during wakefulness.

Breathing sounds spectral and higher order statistics changes from wakefulness to sleep in apneic and non-apneic people.

Breathing sounds analysis conveys valuable information in relation to obstructive sleep apnea (OSA) during both sleep and wakefulness. In this study, we investigated whether the breathings sounds spectral and higher order statistics characteristics (HOS) change from wakefulness to sleep, and more importantly whether this change is associated with severity of OSA. Tracheal breathing sounds of 6 individuals with severe OSA and 6 non-OSA individuals during wakefulness and stage 2 of sleep, both in supine position, were used in this study. The sounds were recorded simultaneously with full overnight polysomnography (PSG) assessment. First, the sounds of 5 noise-free breathing cycles were extracted and sequestered into inspiratory and expiratory phase segments manually for each study subject. After normalizing each sound segment to its energy, spectral and HOS features were calculated. Several features including the median bispectral frequency (MBF), spectral bandwidth (BW) and bispectrum Harmonic Mean (HM) were found to change statistically significantly from wakefulness to sleep mostly in severe OSA group but not as much in non-OSA group. The most prominent and consistent change between the two groups of OSA and non-OSA was observed in MBF; it changed from wakefulness to sleep in the two groups in an opposite manner; this observation is congruent with the hypothesis that the upper airway in OSA population has an increased non-homogeneity.

Implementation of a gelatin model to simulate biological activity in the inner ear for electrovestibulography (EVestG) validation.

In this work, a physical model that simulates electrical activity of the inner ear has been developed. The purpose is to evaluate extraction of vestibular field potentials (FPs) in the presence of various sources of noise by a proprietary software algorithm. The ear model is constructed of gelatin as an alternative to human tissue where independently driven electrical sources are placed in gelatin to mimic various biological signals (muscle, cerebral, and vestibular). Components of system noise (recording apparatus generated noise, electrodes, etc) will be naturally superimposed on the recording, hence enables various recording conditions to be simulated. Muscle activity present in the recordings and noise generated from the recording apparatus were found to be the most dominating sources that degrade performance of FP extraction. The model can be used to provide insights towards enhancing the FP detection algorithm under various signal-to-noise ratios.

Diagnosis of Parkinson's disease using electrovestibulography.

In this paper, a new method for diagnosis of Parkinson's disease (PD) based on the analysis of electrovestibulography (EVestG) signals is introduced. EVestG signals are in fact the vestibular response modulated by more cortical brain signals; they are recorded from the ear canal. EVestG data of 20 individuals with PD and 28 healthy controls were adopted from a previous study. The field potentials and their firing pattern in response to whole body tilt stimuli from both left and right ears were extracted. We investigated several statistical and fractal features of the field potentials and also their firing interval histograms followed by one-way analysis of variance to select pairs of features showing the most significant differences between individuals with Parkinson disease and the age-matched controls. Linear discriminant analysis classification was applied to every selected feature using a leave-one-out routine. The result of each feature's classifier was used in a heuristic average voting system to diagnose PD patients. The results show more than 95% accuracy for PD diagnosis. Given that the patients were at different stage of disease, the high accuracy of the results is encouraging for continuing exploration of the EVestG application to PD diagnosis as it may provide a quick and non-invasive screening tool.

Vestibular spontaneous response as a potential signature for Parkinson's disease.

In this paper, we report on a new method for potential diagnosis of Parkinson's Disease (PD) based on the analysis of the spontaneous response of vestibular system recorded by Electrovestibulography (EVestG). EVestG data of 20 individuals with PD and 28 healthy controls were adopted from a previous study. The field potentials and their firing pattern in response to whole body tilt stimuli from both left and right ears were extracted. We investigated several statistical and fractal features of the field potentials and also their firing patterns. One-way analysis of variance (ANOVA) was used to select the features showing the most significant differences between individuals with PD and the age-matched controls. Linear Discriminant analysis classification was applied to every selected feature using a leave-one-out routine. The result of each feature's classifier was used in a heuristic weighted average voting system to diagnose PD patients. The weights of the voting system were the (posterior) probabilities calculated by the designed classifier to indicate a subject related to a specific class. The results show more than 97% accuracy for PD diagnosis. Given that the patients were at different stage of disease, the high accuracy of the results encourages the use of vestibular response for PD diagnosis as a plausible quick and non-invasive screening tool.

Investigating the statistical properties of the swallowing sounds.

In this paper the statistical properties of the swallowing sound is discussed. This knowledge is required for the acoustical modeling of the swallowing mechanism as it is important to select an appropriate type of the system (i.e. linear vs. nonlinear) for modeling. The tests of linearity and gaussianity were performed. The results of the statistical test of gaussianity showed a nonGaussian distribution of the swallowing sound signals. Also, the test of linearity exhibited the nonlinear characteristics of the model that represents the swallowing sound generation.

Detailed analysis of the relationship between tracheal breath sounds and airflow in relation to OSA during wake and sleep.

Tracheal respiratory sound analysis is a simple, inexpensive and non-invasive way to study the pathology of the upper airways. Recently, it has attracted considerable attention for acoustical flow estimation and investigation of obstruction in the upper airways. Obstructive sleep apena (OSA) is characterized by periods of reduction or complete cessation of airflow during sleep. However, the flow-sound relationship is highly variable among OSA and non-OSA individuals; it also changes for the same person at different body postures and during wake and sleep. In this study we recorded respiratory sound and flow from 93 non-OSA individuals as well as 13 OSA patients during sleep and wake. We investigated the statistical correlation between the flow-sound model parameters and anthropometric features in the non-OSA group. The results have shown that gender, height and smoking are the most significant factors that affect the model parameters. We compared the flow-sound relationship in OSA and non-OSA groups in the sitting position while awake. We also examined the variations in the model parameters in OSA patients during sleep and wake in the recumbent position. The results show that the model parameters are different in the two groups even when accounted for height, gender and position. In OSA group, the model parameters change from wake to sleep, even at the same position. The variations in the model parameters can be used to investigate the characteristics of upper airways and examine the factors that can lead to the upper airways obstruction during sleep.

Application of fractal dimension on vestibular response signals for diagnosis of Parkinson's disease.

In this paper, a novel method based on analysis of dynamic response of vestibular system for diagnosis of Parkinson's Disease (PD) is introduced. Electrovestibulography (EVestG) signals are recorded from the ear canal in response to a vestibular stimulus. EVestG signals are in fact the vestibular response modulated by more cortical brain signals. We used EVestG data of 20 patients with PD and 26 age-matched healthy controls recorded in a previous study. We calculated the Katz Fractal Dimension (FD) of the extracted timing signal of firings during contralateral and ipsilateral stimuli of both left and right ear. We used multivariate analysis of variance (MANOVA) to select pairs of features showing the most significant differences between the groups. Then, Linear and Quadratic Discriminant (LDA, QDA) classification algorithms were applied on the selected features. The results have shown above 77.27% accuracy. Given the small population of the subjects and the patients were at different stage of disease, the results encourage continuing exploration of the application of EVestG for PD diagnosis and perhaps as a quick and non-invasive screening tool.

Heart sound cancellation from lung sound recordings using time-frequency filtering.

During lung sound recordings, heart sounds (HS) interfere with clinical interpretation of lung sounds over the low frequency components which is significant especially at low flow rates. Hence, it is desirable to cancel the effect of HS on lung sound records. In this paper, a novel HS cancellation method is presented. This method first localizes HS segments using multiresolution decomposition of the wavelet transform coefficients, then removes those segments from the original lung sound record and estimates the missing data via a 2D interpolation in the time-frequency (TF) domain. Finally, the signal is reconstructed into the time domain. To evaluate the efficiency of the TF filtering, the average power spectral density (PSD) of the original lung sound segments with and without HS over four frequency bands from 20 to 300 Hz were calculated and compared with the average PSD of the filtered signals. Statistical tests show that there is no significant difference between the average PSD of the HS-free original lung sounds and the TF-filtered signal for all frequency bands at both low and medium flow rates. It was found that the proposed method successfully removes HS from lung sound signals while preserving the original fundamental components of the lung sounds.

Center of mass approximation during walking as a function of trunk and swing leg acceleration.

The 3D center of body mass (COM) trajectory provides us with a measure of movement performance and level of stability while walking. As an alternative to directly calculating the COM from motion trajectories and anthropometric data, we propose developing models to estimate the COM trajectory during walking on irregular surfaces. The inputs to the models were acquired via two accelerometers, one representing the trunk segment placed on T2 and the second representing the swing leg placed on the lateral malleolus. The subjects walked on a fixed surface and encountered an uneven, irregular surface, causing instability in the balance system. The results were encouraging, providing an estimate of the COM trajectory with a low error of 4.17 +/- 1.94%. The reasonable accuracy, portability, ease of use and low cost (compared with video motion analysis systems) of the accelerometers increases the range of clinical applications of the proposed method.

Estimation of the center of body mass during forward stepping using body acceleration.

The center of body mass (COM) calculation is a key factor in analyzing human postural control. In recent years there have been attempts to estimate the COM by body acceleration during standing. In this study, a parabolic model is used to estimate the COM trajectory during forward stepping, using the body segments' accelerations, which were measured by inexpensive and portable accelerometers placed on the trunk and swing leg. Paced and voluntary forward stepping was performed on different support surfaces and with different speeds of stepping. Forward steps were extracted by analyzing the ankle marker position in vertical direction. The model was calibrated by genetic algorithm and tested on forward stepping data using the leave one out method. The results are encouraging for the use of the proposed model as a mean to estimate the COM trajectory during forward stepping.

Adaptive compression of respiratory and swallowing sounds.

Compression of biomedical signals is of great interest in telemedicine applications for a fast and reliable online data transfer. In this paper, an adaptive method for compression of respiratory and swallowing sounds is proposed. The method is based on the transform coding of the signal and adaptively assigns two different bit allocations for encoding the stationary and non-stationary parts of the signal. The method was applied to data of 12 subjects and its performance was compared with other methods of bit allocation such as constant bit allocation. The results show that the proposed adaptive bit allocation method improves the SNR of the signal by 3 dB compared to that of constant bit allocation method. Furthermore, the possibility of removing the requirement of updating the bit allocation method for each subject was investigated. The results confirm that with a few training data, the proposed method can be used in a fully automated mode without the need to adjust the adaptive bit allocation for every signal separately; hence, it is faster for online coding applications.

Respiratory flow estimation from tracheal sound by adaptive filters.

In this study, average power of tracheal sound (Pave) was used to estimate flow by parametric method as well as adaptive filters as a nonparametric method. Based on some preliminary studies, an exponential model was used for describing the relationship between flow and Pave for parametric method. It was assumed that flow signal of at least one breath from each target flow is available for calibration. The error for flow estimation with parametric method, was found to be 9 ± 3 % and 10 ± 4 % for inspiration and expiration, respectively. Considering nonparametric method, the estimation error was the least for the third order adaptive filter using the average power of the tracheal sound (dB), which was 10 ± 3 % and 11 ± 4 % for inspiration and expiration, respectively.

Application of nonlinear dynamics to human postural control system.

Instability, fear of falling and fall injuries are common problems for the older population and individuals with chronic disabilities. Development of screening tools and outcome measures of balance performance and fall risk has become of great interest for health professionals and researchers in the field of rehabilitation. Although linear modeling is desirable due to its simplicity, however, most physiological systems are too complicated to lend themselves to linear analysis because the human body is multi-segmental, and there are feedforward and feedback control schemes of mutual relationship between the segments. In this study, we investigated the use of nonlinear dynamic tools to extract characteristic features of postural sway. Among various methods of nonlinear dynamics, the Rényi fractal dimension and the Rényi spectrum are presented as quantitative descriptors. The center of foot pressure (COP) was recorded during four quiet standing tasks of increasing difficulty (eliminating vision and use of a compliant surface). The COP trajectories' Rényi dimension and Rényi spectra of young healthy subjects with no history of neurological disorder were investigated in comparison to that of elderly patients with balance impairment and history of frequent falls. Sway path, the common linear parameter of postural sway, was also used to investigate its distinction from non-linear parameters. The results of this study suggest the COP trajectories' Rényi dimension and sway path can be used as classifiers for balance disorders, and they provided different indications of postural control system characteristics between the two groups and different task demands.

Development of an interactive motivating tool for rehabilitation movements.

In this paper, an interactive tool, including three computer games controlled via the center of foot pressure (COP) trajectory biofeedback, was designed to aid in pressure balance for rehabilitating persons with balance disorders. The games interact in real-time with the Vista Medical Force Sensitive Applications software and pressure mat. The main goal of this research was to employ attractive and motivational learning techniques, using equipment that is available to a large population, to increase volume of exercise practice and to retain the patient's attention. Questionnaires regarding the motivational aspects of the games were administered to 15 subjects (7 patients). The results indicate that the tools were indeed attractive, motivational and an improvement to conventional exercise regimes.