Search

We present a random forest (RF) classification and regression technique to predict, intraoperatively, the unified Parkinson's disease rating scale (UPDRS) improvement after deep brain stimulation (DBS). We hypothesized that a data-informed combination of features extracted from intraoperative microelectrode recordings (MERs) can predict the motor improvement of Parkinson's disease patients undergoing DBS surgery. We modified the employed RFs to account for unbalanced datasets and multiple observations per patient, and showed, for the first time, that only five neurophysiologically interpretable MER signal features are sufficient for predicting UPDRS improvement. This finding suggests that subthalamic nucleus (STN) electrophysiological signal characteristics are strongly correlated to the extent of motor behavior improvement observed in STN-DBS.

The derivation and application of a method designed to investigate the focusing properties of pulsed baseband signals of short pulsewidth (/spl sim/1 ns) in biological tissue media are reported. To this end, sources fed from TEM waveguides, concentrically placed at the periphery of a three-layer cylindrical lossy model, are assumed. A Fourier-series-based methodology, appropriate for a useful class of pulse train incident signals, is presented and utilized to study the dynamics of pulse propagation inside the tissue medium. The propagation of each spectral component of the incident field within the tissue medium is analyzed by applying an integral-equation technique and a Fourier-series representation is used in order to obtain the time dependence of the electromagnetic fields produced at any point within tissue due to the pulsed excitation of the array elements. Numerical results are computed and presented at several points in a three-layer geometry, 20 cm in diameter, irradiated by an eight-element waveguide array. Focusing at a specific point of interest within tissue is achieved by properly adjusting the time delay of the signals injected to the individual applicators of the array.

Efficient management of the asymptomatic carotid disease remains a crucial challenge in clinical practice, because the ultrasonographically estimated degree of stenosis, which is currently used to determine treatment decisions, has been shown to be inadequate. In this study, texture (morphological) characteristics were investigated in a sample of asymptomatic male subjects, at the atheromatous plaque, the adjacent arterial wall and the plaque shoulder, i.e. the boundary between plaque and adjacent wall. A total of 25 arteries were interrogated, 11 with low (50-69%) and 14 with high (70-100%) degrees of stenosis. The two groups had similar ages. Texture characteristics were estimated from systolic and diastolic B-mode ultrasound images, and included four second-order statistical parameters (contrast, correlation, energy and homogeneity), each calculated at four different image directions (0 degrees, 45 degrees, 90 degrees, 135 degrees), yielding a total of 16 characteristics. Between high and low stenosis groups, 8 out of 16 characteristics were statistically different at the plaque shoulder at systole and 6 at diastole. No differences were observed between the two groups for any of the texture characteristics at the plaque nor at the adjacent wall. Differences in morphology along the arterial wall (wall - shoulder - plaque) were more pronounced in cases of high stenosis. The findings indicated that (a) the plaque shoulder is a particular area, requiring additional investigation so as to better understand the pathophysiology of atherosclerosis, (b) the phase of the cardiac cycle (systole or diastole) is important in texture analysis, and (c) the variability of morphology along the arterial wall, which is indicative of areas of tissue discontinuities, and therefore more vulnerable to rupture, can be characterized quantitatively with texture indices, toward an improved assessment of cardiovascular risk. It can be concluded that ultrasound-based texture indices may reveal novel markers for early detection and monitoring of subjects at high risk of cerebrovascular events, in the context of individualized, noninvasive and affordable diagnosis.

Efficient management of asymptomatic carotid atherosclerosis remains a crucial issue in clinical practice, because the currently used determinant for deciding surgical or medical treatment, namely the stenosis degree, has proven inadequate. Various features derived from ultrasound image analysis have shown promise for enhancing the effectiveness of decision making. Among these, multiresolution, i.e. wavelet-transform-based, texture analysis has been shown to adequately characterise atheromatous tissue, especially in terms of horizontally oriented texture in longitudinal B-mode sections. We sought to investigate the horizontally-oriented multiresolution textural properties of arterial wall and plaque in a carefully selected sample of asymptomatic male subjects. Of the 19 carotid arteries that were interrogated, 7 had low stenosis degrees (50-69%) and 12 had high stenosis degrees (70-100%); the two groups had similar ages. Multiresolution analysis was performed using wavelet packets and the coiflet 1 wavelet, for three levels of decomposition. Seven features were derived from the resulting approximation (A) and horizontal detail (Dh) images, for two segments in each artery, namely plaque and the adjacent intima-media complex (IM). All features were statistically similar between low and high stenosis cases, suggesting that stenosis degree may not be sufficient to characterise the two case types. Differences between IM and plaque were observed for most multiresolution features. The differences were more prominent for higher decomposition levels. The feature Dh(3)A(2)A(1) was different between plaque and IM in the low stenosis cases but was not different in the high stenosis cases. The opposite was observed for Dh(2)Dh(1). Such differences suggest discontinuities in tissue properties and therefore increased vulnerability. In conclusion, multiresolution texture analysis may provide valuable information about normal and atheromatous tissue, because certain tissue properties are revealed at high decomposition levels.

Extracellular recordings in the area of the subthalamic nucleus (STN) of Parkinson's disease patients undergoing deep brain stimulation comprise fast events, Action Potentials and slower events, known as Local Field Potentials (LFP). The LFP is believed to represent the synchronized input into the observed area, as opposed to the spike data, which represents the output. We have shown before that there is an input-output relationship between these two components in the STN. In the present paper, we extend these observations by using LFP-driven Volterra models and the Laguerre expansion technique to estimate nonlinear dynamic models which are able to predict the recorded spiking activity. To this end, we rigorously examine the optimal model order. The improved performance of the second-order Volterra models indicates that there is a nonlinear relationship between the LFP and the spiking activity. To obtain a more compact and readily interpretable model, the most significant dynamic components of the identified Volterra models are extracted using principal dynamic mode analysis.

Ieee Transactions on Information Technology in Biomedicine

Ieee Transactions on Information Technology in Biomedicine

4 1984 1987

414 416