Miodrag Popovic

Real-Time Analysis of EEG Indexes of Alertness, Cognition, and Memory Acquired With a Wireless EEG Headset

The integration of brain monitoring into the man-machine interface holds great promise for real-time assessment of operator status and intelligent allocation of tasks between machines and humans. This article presents an integrated hardware and software solution for acquisition and real-time analysis of the electroencephalogram (EEG) to monitor indexes of alertness, cognition, and memory. Three experimental paradigms were evaluated in a total of 45 participants to identify EEG indexes associated with changes in cognitive workload: the Warship Commander Task (WCT), a simulated navy command and control environment that allowed workload levels to be systematically manipulated; a cognitive task with three levels of difficulty and consistent sensory inputs and motor outputs; and a multisession image learning and recognition memory test. Across tasks and participants, specific changes in the EEG were identified that were reliably associated with levels of cognitive workload. The EEG indexes were also shown to change as a function of training on the WCT and the learning and memory task. Future applications of the system to augment cognition in military and industrial environments are discussed.

Characterization of visually similar diffuse diseases from B-scan liver images using nonseparable wavelet transform

This paper describes a new approach for texture characterization, based on nonseparable wavelet decomposition, and its application for the discrimination of visually similar diffuse diseases of liver. The proposed feature-extraction algorithm applies nonseparable quincunx wavelet transform and uses energies of the transformed regions to characterize textures. Classification experiments on a set of three different tissue types show that the scale/frequency approach, particularly one based on the nonseparable wavelet transform, could be a reliable method for a texture characterization and analysis of B-scan liver images. Comparison between the quincunx and the traditional wavelet decomposition suggests that the quincunx transform is more appropriate for characterization of noisy data, and practical applications, requiring description with lower rotational sensitivity.

Automatic segmentation of intravascular ultrasound images: A texture-based approach

Extraction of blood vessel boundaries from intravascular ultrasound images is essential in the quantitative analysis of cardiovascular functions. In this study, we are presenting a completely automated procedure for determining blood vessel borders. This approach uses textural operators to separate different tissue regions and morphological processing to refine extracted contours. The method was tested in a set of 29 intravascular ultrasound images obtainedin vivo. To assess the performance of the method, we have compared the automatically processed images with the manual tracings, using three different criteria: correlation coefficient, match ratio, and relative error of computed shape parameters. In both contour detection and shape parameters estimation, the proposed method yielded consistently good results. Due to its robustness and accuracy, this approach is appropriate for clinical use, whereas computational efficiency of the method facilitates low-cost implementation.

On the selection of an optimal wavelet basis for texture characterization

Many issues related to the choice of filter bank in the texture processing remained unresolved till now. The impact of the wavelet basis has been mentioned in a very few of the previously published papers whereas other more detailed investigations have considered only the choice of the wavelet basis in image coding. Therefore, the scope of this paper was to investigate whether the properties of the decomposition filters play an important role in texture description, and which feature is dominant in the selection of an optimal filter bank. We performed classification experiments with 23 Brodatz textures and ranked 19 orthogonal and biorthogonal filters. Our experiments show that the selection of the decomposition filters has a significant influence on the texture characterization. Based on the obtained results, we establish the most relevant criteria for choice of decomposition filters in wavelet-based texture characterization algorithms.

Characterization of visually similar diffuse diseases from B-scan liver images with the nonseparable wavelet transform

This paper describes the application of the nonseparable wavelet decomposition for the discrimination of diffuse diseases of liver. The proposed feature extraction algorithm uses the filter bank performing the quincunx transform and characterizes textures by a set of channel variances estimated at the output of each filter. Classification experiments on a set of three different tissue types show that this approach could be a reliable method for analysis of B-scan liver images.

Myocardial Tissue Characterization After Acute Myocardial Infarction With Wavelet Image Decomposition A Novel Approach for the Detection of Myocardial Viability in the Early Postinfarction Period

BACKGROUND Only a few texture measures can be used for texture characterization of infarcted myocardium and detection of reperfused myocardium early after infarction. This study was conducted to establish the relationship between texture properties of infarcted myocardium and infarct-related artery patency by quantitative computer analysis of 2-dimensional echocardiographic images with the wavelet-based method for texture characterization, evaluate the relationship between texture properties and myocardial viability, and correlate histopathologic changes after experimental infarction with the texture measures. METHODS AND RESULTS We analyzed 2-dimensional transthoracic echocardiographic images in 18 patients at different time points after infarction using the wavelet transform method. Regional wall motion of infarcted segments was analyzed on a follow-up echocardiographic study obtained 6 months after infarction. To verify the accuracy of the proposed texture measure and energy difference cutoff value, we prospectively evaluated another group of 19 patients. In addition, histopathologic changes in 9 dogs with experimental infarction were correlated with the texture measures. Sensitivity, specificity, and accuracy of the wavelet method for detection of reperfusion in the study group were 73%, 86%, and 78%, respectively, on day 2; 91%, 86%, and 89%, at 1 week; and 100%, 100%, and 100% at 3 weeks. Among 9 patients with improvement in regional wall motion on a follow-up study, 7 on day 2, 8 at 1 week, and 9 at 3 weeks were classified into the reperfused group by the wavelet method. Histopathologic features associated with the classification of reperfusion by the wavelet method were infarct transmurality (P=0.024) and degree of necrosis (P=0.028). CONCLUSIONS Our clinical and experimental data suggest that the wavelet method can be used to differentiate between viable myocardium with recovery potential and definite myocardial necrosis in the early postinfarction period.

EEG quantification of alertness: Methods for early identification of individuals most susceptible to sleep deprivation

Electroencephalographic (EEG) and neurocognitive measures were simultaneously acquired to quantify alertness from 24 participants during 44-hours of sleep deprivation. Performance on a three-choice vigilance task (3C-VT), paired-associate learning/memory task (PAL) and modified Maintenance of Wakefulness Test (MWT), and sleep technician-observed drowsiness (eye-closures, head-nods, EEG slowing) were quantified. The B-Alert system automatically classifies each second of EEG on an alertness/drowsiness continuum. B-Alert classifications were significantly correlated with technician-observations, visually scored EEG and performance measures. B-Alert classifications during 3C-VT, and technician observations and performance during the 3C-VT and PAL evidenced progressively increasing drowsiness as a result of sleep deprivation with a stabilizing effect observed at the batteries occurring between 0600 and 1100 suggesting a possible circadian effect similar to those reported in previous sleep deprivation studies. Participants were given an opportunity to take a 40-minute nap approximately 24-hours into the sleep deprivation portion of the study (i.e., 7 PM on Saturday). The nap was followed by a transient period of increased alertness. Approximately 8 hours after the nap, behavioral and physiological measures of drowsiness returned to levels prior to the nap. Cluster analysis was used to stratify individuals into three groups based on their level of impairment as a result of sleep deprivation. The combination of B-Alert and neuro-behavioral measures may identify individuals whose performance is most susceptible to sleep deprivation. These objective measures could be applied in an operational setting to provide a “biobehavioral assay” to determine vulnerability to sleep deprivation.

A new look at the comparison of the fast Hartley and Fourier transforms

In the correspondence, a fair comparative analysis of algorithms for the fast Hartley transform (FHT) and the real valued fast Fourier transform (RVFFT) is presented. The complexity analysis and run-time comparisons are conducted and explained simultaneously. The complexity analysis shows great similarity between RVFFT and FHT. The run-time comparisons also show negligible differences, with some small advantages to RVFFT. The influence of the compiler on the execution time may be more significant than the choice of the algorithm. Also, a new and more accurate model for the prediction of execution time of the algorithm is proposed. >

A new improvement to the Powell and Chau linear phase IIR filters

An improvement to the realization of the linear-phase IIR filters is described. It is based on the rearrangement of the numerator polynomials of the IIR filter functions that are used in the real-time realizations proposed in literature. The new realization has better total harmonic distortion when a sine input is used, and it has smaller phase and group delay errors due to finite section length.

Classification of the ultrasound liver images with the 2N/spl times/1-D wavelet transform

The authors propose a new separable extension of the 1-D wavelet transform to the 2-D case and describe its application to the texture characterization problem. Comparing to previous decompositions, with the same resolution levels for each subband in horizontal and vertical directions, the new method has different resolutions for different directions. The new algorithm is applied to 84 ultrasound liver images, to detect liver cirrhosis in its early stage. The classification accuracy was 92%. The method was compared to other texture description methods (gray level cooccurrence, Laws filters, pyramid and tree structured wavelet decompositions). The proposed 2N/spl times/1-D wavelet decomposition, gave the highest classification rate, showing its applicability for the approach based analysis of the large class of natural textures.