Salvador Olmos

A wavelet-based ECG delineator: evaluation on standard databases

In this paper, we developed and evaluated a robust single-lead electrocardiogram (ECG) delineation system based on the wavelet transform (WT). In a first step, QRS complexes are detected. Then, each QRS is delineated by detecting and identifying the peaks of the individual waves, as well as the complex onset and end. Finally, the determination of P and T wave peaks, onsets and ends is performed. We evaluated the algorithm on several manually annotated databases, such as MIT-BIH Arrhythmia, QT, European ST-T and CSE databases, developed for validation purposes. The QRS detector obtained a sensitivity of Se=99.66% and a positive predictivity of P+=99.56% over the first lead of the validation databases (more than 980,000 beats), while for the well-known MIT-BIH Arrhythmia Database, Se and P+ over 99.8% were attained. As for the delineation of the ECG waves, the mean and standard deviation of the differences between the automatic and manual annotations were computed. The mean error obtained with the WT approach was found not to exceed one sampling interval, while the standard deviations were around the accepted tolerances between expert physicians, outperforming the results of other well known algorithms, especially in determining the end of T wave.

Methodological principles of T wave alternans analysis: a unified framework

Visible T wave alternans (TWA) in the electrocardiogram (ECG) had been regarded as an infrequent phenomenon during the first 80 years of electrocardiography. Nevertheless, computerized analysis changed this perception. In the last two decades, a variety of techniques for automatic TWA analysis have been proposed. These techniques have allowed researchers to detect nonvisible TWA in a wide variety of clinical and experimental conditions. Such studies have recently shown that TWA is related to cardiac instability and increased arrhythmogenicity. Comparison of TWA analysis methods is a difficult task due to the diversity of approaches. In this paper, we propose a unified framework which holds the existing methods. In the light of this framework, the methodological principles of the published TWA analysis schemes are compared and discussed. This framework may have an important role to develop new approaches to this problem.

Adaptive estimation of QRS complex wave features of ECG signal by the Hermite model

The most characteristic wave set in ECG signals is the QRS complex. Automatic procedures to classify the QRS are very useful in the diagnosis of cardiac dysfunctions. Early detection and classification of QRS changes are important in realtime monitoring. ECG data compression is also important for storage and data transmission. An Adaptive Hermite Model Estimation System (AHMES) is presented for on-line beat-to-beat estimation of the features that describe the QRS complex with the Hermite model. The AHMES is based on the multiple-input adaptive linear combiner, using as inputs the succession of the QRS complexes and the Hermite functions, where a procedure has been incorporated to adaptively estimate a width related parameter b. The system allows an efficient real-time parameter extraction for classification and data compression. The performance of the AHMES is compared with that of direct feature estimation, studying the improvement in signal-to-noise ratio. In addition, the effect of misalignment at the QRS mark is shown to become a neglecting low-pass effect. The results allow the conditions in which the AHMES improves the direct estimate to be established. The application is shown, for subsequent classification, of the AHMES in extracting the QRS features of an ECG signal with the bigeminy phenomena. Another application is highlighted that helps wide ectopic beats detection using the width parameter b.

Characterization of repolarization alternans during ischemia: time-course and spatial analysis

T-wave alternans (TWA) has been linked to increased vulnerability to ventricular fibrillation in different settings including myocardial ischemia. In this study, we propose a methodology for the characterization of TWA induced by transient, regional ischemia. We studied the prevalence, magnitude and spatio-temporal relationship between TWA and ischemia in 95 patients undergoing percutaneous transluminal coronary angioplasty (PTCA). Two electrocardiogram records of each patient, a control recording before PTCA and the PTCA record, were analyzed using a robust, recently proposed method for TWA analysis. The detected episodes were characterized in terms of their time-course, lead distribution and alternans waveform. One third of the patients (33.7%) showed TWA episodes during PTCA. The highest prevalence (51.7%) and amplitude were found in patients with left anterior descendent artery occlusion. The onset of TWA was detected after the first 1-2 min of occlusion, suggesting that some level of ischemia must be attained before TWA arises, while disappearance of TWA following reperfusion was much more rapid. The TWA lead distributions and waveforms showed distinct distributions according to the occluded artery reflecting the regional nature of the TWA phenomenon.

Automatic detection of ST-T complex changes on the ECG using filtered RMS difference series: application to ambulatory ischemia monitoring

A new detector is presented which finds changes in the repolarization phase (ST-T complex) of the cardiac cycle. It operates by applying a detection algorithm to the filtered root mean square (rms) series of differences between the beat segment (ST segment or ST-T complex) and an average pattern segment. The detector has been validated using the European ST-T database, which contains ST-T complex episodes manually annotated by cardiologists, resulting in sensitivity/positive predictivity of 85/86%, and 85/76%, for ST segment deviations and ST-T complex changes, respectively. The proposed detector has a performance similar to those which have a more complicated structure. The detector has the advantage of finding both ST segment deviations and entire ST-T complex changes thereby providing a wider characterization of the potential ischemic events. A post-processing stage, based on a cross-correlation analysis for the episodes in the rms series, is presented. With this stage subclinical events with repetitive pattern were found in around 20% of the recordings and improved the performance to 90/85%, and 89/76%, for ST segment and ST-T complex changes, respectively.

Registration of Anatomical Images Using Paths of Diffeomorphisms Parameterized with Stationary Vector Field Flows

Computational Anatomy aims for the study of variability in anatomical structures from images. Variability is encoded by the spatial transformations existing between anatomical images and a template selected as reference. In the absence of a more justified model for inter-subject variability, transformations are considered to belong to a convenient family of diffeomorphisms which provides a suitable mathematical setting for the analysis of anatomical variability. One of the proposed paradigms for diffeomorphic registration is the Large Deformation Diffeomorphic Metric Mapping (LDDMM). In this framework, transformations are characterized as end points of paths parameterized by time-varying flows of vector fields defined on the tangent space of a Riemannian manifold of diffeomorphisms and computed from the solution of the non-stationary transport equation associated to these flows. With this characterization, optimization in LDDMM is performed on the space of non-stationary vector field flows resulting into a time and memory consuming algorithm. Recently, an alternative characterization of paths of diffeomorphisms based on constant-time flows of vector fields has been proposed in the literature. With this parameterization, diffeomorphisms constitute solutions of stationary ODEs. In this article, the stationary parameterization is included for diffeomorphic registration in the LDDMM framework. We formulate the variational problem related to this registration scenario and derive the associated Euler-Lagrange equations. Moreover, the performance of the non-stationary vs the stationary parameterizations in real and simulated 3D-MRI brain datasets is evaluated. Compared to the non-stationary parameterization, our proposal provides similar results in terms of image matching and local differences between the diffeomorphic transformations while drastically reducing memory and time requirements.

Evaluation of a wavelet-based ECG waveform detector on the QT database

We have evaluated a single-lead wavelet transform (WT) based detector of ECG significant points. A quadratic spline wavelet was used as prototype wavelet, and the first four scales of the Dyadic WT were analyzed. First of all, we detect QRS complexes. Then, the individual waves, the onset and the offset of the QRS complexes are identified and finally P and T peaks and their onset and offset are detected. We have validated the algorithm with the manual annotations in the QT Database (QTDB), developed for validation purposes. QRS and other ECG waveform boundaries were independently evaluated. The mean and standard deviation of the differences between the manual and detectors wave boundary annotations were calculated. The standard deviations obtained with the WT approach are around the accepted tolerances between expert physicians, outperforming the results of a low-pass differentiator algorithm, which was used as a reference, especially in the T wave offset. The QRS detector obtained a sensitivity of Se=99.91 and a positive predictivity of P+=99.88%.

Block adaptive filters with deterministic reference inputs for event-related signals: BLMS and BRLS

Adaptive estimation of the linear coefficient vector in truncated expansions is considered for the purpose of modeling noisy, recurrent signals. Two different criteria are studied for block-wise processing of the signal: the mean square error (MSE) and the least squares (LS) error. The block LMS (BLMS) algorithm, being the solution of the steepest descent strategy for minimizing the MSE, is shown to be steady-state unbiased and with a lower variance than the LMS algorithm. It is demonstrated that BLMS is equivalent to an exponential averager in the subspace spanned by the truncated set of basis functions. The block recursive least squares (BRLS) solution is shown to be equivalent to the BLMS algorithm with a decreasing step size. The BRLS is unbiased at any occurrence number of the signal and has the same steady-state variance as the BLMS but with a lower variance at the transient stage. The estimation methods can be interpreted in terms of linear, time-variant filtering. The performance of the methods is studied on an ECG signal, and the results show that the performance of the block algorithms is superior to that of the LMS algorithm. In addition, measurements with clinical interest are found to be more robustly estimated in noisy signals.

Malignancy assessment of brain tumours with magnetic resonance spectroscopy and dynamic susceptibility contrast MRI

Magnetic resonance imaging (MRI) is the most common and well-established imaging modality for evaluation of intracerebral neoplasms, but there are still some incompletely solved challenges, such as reliable distinction between high- and low-grade tumours, exact delineation of tumour extension, and discrimination between recurrent tumour and radiation necrosis. The aim of this study was to evaluate the contribution of two MRI techniques to non-invasively estimate brain tumour grade. Twenty-four patients referred to MRI examination were analyzed and diagnosed with single intra-axial brain tumour. Lastly, histopathological analysis was performed to verify tumour type. Ten patients presented low-grade gliomas, while the remaining patients showed high-grade tumours, including glioblastomas in eight cases, isolated metastases in four patients and two cases with anaplastic gliomas. MRI examinations were performed on a 1.5-T scanner (Signa, General Electric). The acquisition protocol included the following sequences: saggital T1-weighted localizer, axial T1- and T2-weighted MRI, single-voxel magnetic resonance spectroscopy (MRS), dynamic susceptibility contrast (DSC) MRI and contrast-enhanced T1-weighted MRI. MRS data was analyzed with standard software provided by the scanner manufacturer. The metabolite ratio with the largest significant difference between tumour grades was the choline/creatine (Ch/Cr) ratio with elevated values in high-grade gliomas and metastases. A Ch/Cr ratio equal or larger than 1.55 predicted malignancy grade with 92% sensitivity and 80% specificity. The area under the ROC curve was 0.92 (CI: 95%; 0.81-1). Regarding to perfusion parameters, relative cerebral blood volume (rCBV) maps were estimated from the MR signal intensity time series during bolus passage with two commercial software packages. Two different regions of interest (ROI) were used to evaluate rCBV: lesion centre and perilesional region. All rCBV values were normalized to CBV in a contrallateral normal appearing white matter region. Statistical differences were not found between different tumour types. However, the presence of blood-brain barrier (BBB) damage was illustrated from concentration-time curves calculated in DSC-MRI. A cluster analysis of the time series was used to identify regions with contrast agent extravasation where T1-effects are superimposed to T2-effects. The presence of BBB damage from concentration-time curves was highly correlated with enhancement of post-contrast T1-weighted images and predicted tumour malignancy with a 92% sensitivity and 90% specificity. A large spatial heterogeneity in concentration-time curves was observed from the cluster analysis, supporting the assumption that ROI selection to compute hemodynamic parameters must be done carefully in order to extract robust parameters.

Contributions to 3D diffeomorphic atlas estimation: application to brain images

This paper focuses on the estimation of statistical atlases of 3D images by means of diffeomorphic transformations. Within a Log-Euclidean framework, the exponential and logarithm maps of diffeomorphisms need to be computed. In this framework, the Inverse Scaling and Squaring (ISS) method has been recently extended for the computation of the logarithm map, which is one of the most time demanding stages. In this work we propose to apply the Baker-Campbell-Hausdorff (BCH) formula instead. In a 3D simulation study, BCH formula and ISS method obtained similar accuracy but BCH formula was more than 100 times faster. This approach allowed us to estimate a 3D statistical brain atlas in a reasonable time, including the average and the modes of variation. Details for the computation of the modes of variation in the Sobolev tangent space of diffeomorphisms are also provided.