Nikolaos Maglaveras

Hybrid image segmentation using watersheds and fast region merging

A hybrid multidimensional image segmentation algorithm is proposed, which combines edge and region-based techniques through the morphological algorithm of watersheds. An edge-preserving statistical noise reduction approach is used as a preprocessing stage in order to compute an accurate estimate of the image gradient. Then, an initial partitioning of the image into primitive regions is produced by applying the watershed transform on the image gradient magnitude. This initial segmentation is the input to a computationally efficient hierarchical (bottom-up) region merging process that produces the final segmentation. The latter process uses the region adjacency graph (RAG) representation of the image regions. At each step, the most similar pair of regions is determined (minimum cost RAG edge), the regions are merged and the RAG is updated. Traditionally, the above is implemented by storing all RAG edges in a priority queue. We propose a significantly faster algorithm, which additionally maintains the so-called nearest neighbor graph, due to which the priority queue size and processing time are drastically reduced. The final segmentation provides, due to the RAG, one-pixel wide, closed, and accurately localized contours/surfaces. Experimental results obtained with two-dimensional/three-dimensional (2-D/3-D) magnetic resonance images are presented.

Watershed-based image segmentation with fast region merging

A hybrid image segmentation algorithm is proposed which combines edge- and region-based techniques through the morphological algorithm of watersheds and consists of the following steps: (a) edge-preserving noise reduction, (b) gradient approximation, (c) detection of watersheds on gradient magnitude image, and (d) hierarchical region merging (HRM) in order to get semantically meaningful segmentations. HRM uses the region adjacency graph (RAG) representation of the image regions. At each step, the most similar pair of regions is determined (minimum cost RAG edge), the regions are merged and the RAG is updated. Traditionally, the above is implemented by storing all the RAG edges in a priority queue (heap). We propose a significantly faster algorithm which maintains an additional graph, the most similar neighbor graph, through which the priority queue size and processing time are drastically reduced. In addition, this region based representation provides one-pixel wide, closed, and accurately localized contours/surfaces. Experimental results using 2D real images are presented.

Model-based processing scheme for quantitative 4-D cardiac MRI analysis

Presents an integrated model-based processing scheme for cardiac magnetic resonance imaging (MRI), embedded in an interactive computing environment suitable for quantitative cardiac analysis, which provides a set of functions for the extraction, modeling, and visualization of cardiac shape and deformation. The methods apply 4-D processing (three spatial and one temporal) to multiphase multislice MRI acquisitions and produce a continuous 4-D model of the myocardial surface deformation. The model is used to measure diagnostically useful parameters, such as wall motion, myocardial thickening, and myocardial mass measurements. The proposed model-based shape extraction method has the advantage of integrating local information into an overall representation and produces a robust description of cardiac cavities. A learning segmentation process that incorporates a generating-shrinking neural network is combined with a spatiotemporal parametric modeling method through functional basis decomposition. A multiscale approach is adopted, which uses at each step a coarse-scale model defined at the previous step in order to constrain the boundary detection. The main advantages of the proposed methods are efficiency, lack of uncertainty about convergence, and robustness to image artifacts.

A study of heart rate and brain system complexity and their interaction in sleep-deprived subjects

In this work we investigate the synchronization of the dynamic behaviour of heart rate (ECG) and brain (EEG) signals using sample entropy as a measure of complexity. EEG and ECG recordings were collected during experiment with sleep-deprived subjects exposed to real field driving conditions. The degree to which brain and heart complexity loose complexity in a synchronous manner, indicating a possible interaction between the two systems is investigated. Preliminary results obtained from the examination of four subjects show the existence of a weak-to-intermediate cross-correlation between these pairs of biological oscillators. Furthermore, the frequency content in both heart rate and brain signals was calculated via power spectrum analysis and the association of synchronisation patterns with prevalent frequencies in the two systems was investigated.

Can P Wave Wavelet Analysis Predict Atrial Fibrillation After Coronary Artery Bypass Grafting

VASSILIKOS V., et al.: Can P Wave Wavelet Analysis Predict Atrial Fibrillation After Coronary Artery Bypass Grafting? The purpose of this study was the evaluation of Morlet wavelet analysis of the P wave as a means of predicting the development of atrial fibrillation (AF) in patients who undergo coronary artery bypass grafting (CABG). The P wave was analyzed using the Morlet wavelet in 50 patients who underwent successful CABG. Group A consisted of 17 patients, 12 men and 5 women, of mean age 66.9 ± 5.9 years , who developed AF postoperatively. Group B consisted of 33 patients, 29 men and 4 women, mean age 62.4 ± 7.8 years , who remained arrhythmia‐free. Using custom‐designed software, P wave duration and wavelet parameters expressing the mean and maximum energy of the P wave were calculated from 3‐channel digital recordings derived from orthogonal ECG leads (X, Y, and Z), and the vector magnitude (VM) was determined in each of 3 frequency bands (200–160 Hz, 150–100 Hz and 90–50 Hz). Univariate logistic‐regression analysis identified a history of hypertension, the mean and maximum energies in all frequency bands along the Z axis, the mean and maximum energies (expressed by the VM) in the 200–160 Hz frequency band, and the mean energy in the 150–100 Hz frequency band along the Y axis as predictors for post‐CABG AF. Multivariate analysis identified hypertension, ejection fraction, and the maximum energies in the 90–50 Hz frequency band along the Z and composite‐vector axes as independent predictors. This multivariate model had a sensitivity of 91% and a specificity of 65%. We conclude that the Morlet wavelet analysis of the P wave is a very sensitive method of identifying patients who are likely to develop AF after CABG. The occurrence of post‐CABG AF can be explained by a different activation pattern along the Z axis. (PACE 2003; 26[Pt. II]:305–309)

Screening of patients with obstructive sleep Apnea syndrome using C4.5 algorithm based on non linear analysis of respiratory signals during sleep

Aim: To classify patients with possible diagnosis of Obstructive Sleep Apnea Syndrome (OSAS) into groups according to the severity of the disease using a decision tree producing algorithm based on nonlinear analysis of 3 respiratory signals instead of the use of full polysomnography. Patients-Methods: Eighty-six consecutive patients referred to the Sleep Unit of a Pulmonology Department underwent full polysomnography and their tests were manually scored. Three nonlinear indices (Largest Lyapunov Exponent-LLE, Detrended Fluctuation Analysis-DFA and Approximate Entropy-APEN) were extracted from two respiratory signals (nasal cannula flow-F and thoracic belt-T). The oxygen saturation signal (SpO2) was also selected. The above measurements provided data to the C4.5 algorithm using a data mining application. Results: Two decision trees were produced using linear and nonlinear data from 3 respiratory signals. The discrimination between normal subjects and sufferers from OSAS presented an accuracy of 84.9% and a recall of 90.3% using the variables age, sex, DFA from F and Time with SpO2<90% (T90). The classification of patients into severity groups had an accuracy of 74.2% and a recall of 81.1% using the variables APEN from F, DFA from F and T90. Conclusion: It is possible to have reliable predictions of the severity of OSAS using linear and nonlinear indices from only two respiratory signals during sleep instead of performing full polysomnography. The proposed algorithm could be used for screening patients suspected to suffer from OSAS.

Coronary arterial tree extraction based on artery tracking and mathematical morphology

An algorithm for the unsupervised extraction of the coronary arterial tree in single-view angiograms is proposed. Its output is a structural description of the coronary arterial tree (skeleton and borders) along with accurate information for the coronary artery dimensions. The method consists of two stages. (i) Arterial tree detection, where the approximate centerline and borders of the coronary arterial tree are extracted through a recursive artery tracking method based on circular template analysis for the local artery border detection. (ii) Artery skeleton and border estimation, where the accurate skeleton and borders of each artery segment of the arterial tree are computed based on the morphological tools of homotopy modification and watershed transform. Specifically, the approximate centerline and borders of each artery segment computed at the first stage are used for constructing its enclosing area where the defined skeleton and border curves are considered as markers. Experimental results using digitized coronary angiograms are presented.

Detection and modeling of infarcted myocardium regions in MRI images using a contour deformable model

This work provides a semi-automatic method for defining and modeling of the infarcted myocardial tissue in MRI images. A deformable contour model based on Fourier decomposition is used to define the border of the infarcted region in successive slice images. A new fast algorithm has been developed for fitting the curve to the borders of the region. The method includes a two-stage process which detects boundary points and directly calculates the model parameters. The method has been tested on MR images from patients with myocardial infarction. Results show that the infarcted region modeling method performs well, being fast, accurate and relatively insensitive to image noise and inhomogeneities.

QRS analysis using wavelet transformation for the prediction of response to cardiac resynchronization therapy: a prospective pilot study.

BACKGROUND Wider QRS and left bundle branch block morphology are related to response to cardiac resynchronization therapy (CRT). A novel time-frequency analysis of the QRS complex may provide additional information in predicting response to CRT. METHODS Signal-averaged electrocardiograms were prospectively recorded, before CRT, in orthogonal leads and QRS decomposition in three frequency bands was performed using the Morlet wavelet transformation. RESULTS Thirty eight patients (age 65±10years, 31 males) were studied. CRT responders (n=28) had wider baseline QRS compared to non-responders and lower QRS energies in all frequency bands. The combination of QRS duration and mean energy in the high frequency band had the best predicting ability (AUC 0.833, 95%CI 0.705-0.962, p=0.002) followed by the maximum energy in the high frequency band (AUC 0.811, 95%CI 0.663-0.960, p=0.004). CONCLUSIONS Wavelet transformation of the QRS complex is useful in predicting response to CRT.

Preventing obesity and eating disorders through behavioural modifications: The SPLENDID vision

Recent intensive research in the fields of obesity and eating disorders has proved most traditional interventions inadequate: The obesity-targeting interventions have either failed or are strongly social context dependent, while the interventions for eating disorders have poor results and high levels of relapse. On the contrary, recent randomized control trials have illustrated that supervised training of patients to eat and move in a non-pathological way is effective in the prevention of both obesity and eating disorders. Applying the same kind of methodologies to the general population in real life conditions for prevention purposes comes as the logical next step. SPLENDID is a recently initiated EU-funded collaborative project that intends to develop a personalised guidance system for helping and training children and young adults to improve their eating and activity behaviour. By combining expertise in behavioural patterns with current advancements in intelligent systems and sensor technologies, SPLENDID is going to detect subjects at risk for developing obesity or eating disorders and offer them enhanced monitoring and guidance to prevent further disease progression. Both behavioural data collection and system evaluation are going to be performed via pilot studies supported by expert health professionals.