Over the past five years, artificial intelligence (AI) has introduced new models and methods for addressing the challenges associated with the broader adoption of AI models and systems in medicine. This paper reviews recent advances in AI for medical image and video analysis, outlines emerging paradigms, highlights pathways for successful clinical translation, and provides recommendations for future work. Hybrid Convolutional Neural Network (CNN) Transformer architectures now deliver state-of-the-art results in segmentation, classification, reconstruction, synthesis, and registration. Foundation and generative AI models enable the use of transfer learning to smaller datasets with limited ground truth. Federated learning supports privacy-preserving collaboration across institutions. Explainable and trustworthy AI approaches have become essential to foster clinician trust, ensure regulatory compliance, and facilitate ethical deployment. Together, these developments pave the way for integrating AI into radiology, pathology, and wider healthcare workflows.

Emotional recognition is a fundamental task towards the true understanding of the instantaneous brain state. Promising results using electroencephalography (EEG) for emotion detection have been presented, yet most approaches are constrained to laboratory settings with limited ecological validity. To address this, we employed an immersive curved-screen projection environment, chosen for its ability to preserve ecological validity while minimizing EEG signal artifacts commonly induced by virtual reality headsets. We present a novel EEG dynamic functional connectivity (dFC) framework for emotional recognition by using the Neural Gas algorithm for capturing the rapid EEG alterations in an immersive projection environment. EEG responses from 33 participants were collected during video stimulus presentations featuring five emotional environments: calm beach, calm nature, shark attack, rollercoaster, and night walk. The EEG acquisition employed a Unicorn Hybrid Black 8-channel system, while the dFC is computed by using weighted Phase Lag Index (wPLI) through a shifting window-based approach. The symbolic time series are built through Neural Gas, and chronnectomics are then calculated. Statistically significant differences were observed in the brain state flexibility across emotional conditions, with calm conditions showing enhanced flexibility (Nature: 0.93, Beach: 0.93) compared to stress conditions (Shark Attack: 0.75, Night Walk: 0.77). The flexibility index emerged as the primary discriminator of emotional states, with all stress versus calm comparisons achieving statistical significance$(p<0.001)$. Mean dwell time and occupancy entropy provided complementary insights into network stability and state diversity. This advanced EEG DFG study in an immersive real-world environment demonstrates superior discrimination of emotional states compared to traditional laboratory settings, establishing new methodological standards for ecological emotion recognition research.

Accurate classification of lung cancer subtypes from histopathological images is challenging due to limited labelled data and high visual similarity between classes. This limitation is especially critical in clinical diagnostics, where timely and accurate classification can impact treatment decisions. To address this, LungCLR presents a comprehensive evaluation of a two-stage framework that combines contrastive self-supervised learning with EfficientNet-B3 for effective feature extraction. In the first stage, the SimCLR framework is employed to pretrain the encoder on unlabeled histopathology images, enabling it to learn meaningful representations by distinguishing subtle morphological variations. A projection head is incorporated to optimise the NT-Xent loss during training. In the second stage, a lightweight classification head is attached and fine-tuned using small labelled subsets, as few as$\mathbf{1 0 0}$samples per class. Finally, partial end-to-end fine-tuning is applied to further enhance performance. LungCLR is evaluated on the LC25000 dataset, which includes three tissue categories: benign, adenocarcinoma, and squamous cell carcinoma. Using the full dataset, the proposed model achieves an impressive accuracy of 99.97 %, outperforming previous state-of-the-art methods. Importantly, even under low data conditions, the model performs robustly, reaching 90.03% accuracy with only 100 labelled samples per class. These results highlight the effectiveness of established contrastive learning techniques when carefully applied in a clinically relevant, lowdata setting.

PulSense is a portable, low-cost and multi-sensor Cardiovascular monitoring and Arrhythmia detection system. The Movesense Medical (MD) multisensory device is used to capture single-lead electrocardiogram (ECG) signals and is integrated into a Raspberry Pi 4 to perform signal processing and arrhythmia detection for first time. The system features a new user-friendly interface. It employs machine learning by a Convolutional Neural Network (CNN) trained on the MIT-BIH Arrhythmia Database (110,000 multi-labeled heartbeats) to accurately classify arrhythmia types. Feature extraction is enhanced by applying a median filter followed by a notch filter and a Continuous Wavelet Transform (CWT). High overall F1 scores were observed in different classes compared to the literature. Although the PulSense prototype is in a continuous phase of development and testing, the experimental results are encouraging and support its further development into a viable solution for constant heart monitoring and the timely detection of cardiovascular diseases in clinical and non-clinical environments.

The present study explores the effect of multichannel transcranial Direct Current Stimulation (mc-tDCS) on directed EEG connectivity patterns in a patient with refractory focal epilepsy. A double-blind, sham-controlled N -of- 1 trial was conducted, where mc-tDCS was applied over a two-week period, with EEG recordings acquired before and after stimulation and sham procedures accordingly. After artifact reduction on the EEG recordings, Generalized Partial Directed Coherence (gPDC) was utilized, to investigate effective connectivity alterations in the patient's EEG recordings. Machine learning models were also employed to evaluate the connectivity findings and the interictal spike-related class (spike / non-spike) separability. The connectivity analysis demonstrated a significant reduction in gPDC connectivity around the key EEG channels associated with epileptic activity, specifically interictal epileptiform discharges (IEDs), following mc-tDCS, with no significant changes observed in the sham condition. Following feature extraction from the connectivity analysis, machine learning validation supported these findings, revealing a potential decrease in the severity of epileptic activity, as indicated by IEDs. The results suggest that mc-tDCS effectively moderates brain connectivity in refractory focal epilepsy, with implications for reducing the frequency of IEDs. This study highlights the potential of integrating advanced connectivity analysis with machine learning for evaluating mctDCS and similar neuromodulation therapies in epilepsy, particularly in modulating pathological brain network dynamics associated with epileptic discharges.

The detection of brain functional alterations related to Epilepsy is crucial in tackling this condition and administering effective patient care. In the last years, several methods have been proposed for accurate non-invasive temporal detection and characterization of seizures in epilepsy using electroencephalography (EEG) data as input. These methods usually follow machine or deep learning frameworks with the majority of the proposed solutions being supervised. In this study, we propose an unsupervised approach for seizure detection using deep learning-based autoencoder-derived features from EEG data. Our method employs deep learning techniques and Markov Chain modeling to automatically identify subtle alterations in high-frequency EEG temporal dynamics associated with seizures. The openly available CHB-MIT database was used for our evaluations. From the results, we demonstrate the potential of the proposed approach in detecting abnormal dynamic behaviors potentially associated with ictal events, achieving ROC-AUC scores in the range of 84-100%, without depending on extensive labeling efforts. The proposed pipeline potentially contributes towards the automation of detecting seizure onset in epilepsy.

Electroencephalography (EEG)-based classification of brain disease such as epilepsy or schizophrenia, decoding brain activity during movement and vision have been shown promising results in the last years. Here, we introduce a novel pipeline for the presence of speech information carried on EEG signals. The proposed work includes a new conducted EEG dataset of 15 subjects and a deep learning model to predict the colour information. With a unique experimental set up, the data successfully captures the information about the mental enunciation of the set of used colors. The primary goal is to perform multiclass classification using our custom EEG data which records the brain activity of individuals during mental enunciation and thought about a class of objects, in our case, colours. Continuous Wavelet Transform (CWT) is applied on each of the EEG channels of each participant to obtain time-frequency (TF) based characteristics. A Vision Transformer (ViT) based model is then developed and used to capture information from these TF. The method deals with a 6-class classification problem, for which, the 6 different colors are used as target classes for our model. The proposed model achieves 91.36% cross validation accuracy, 5.48x the random guess accuracy. These results clearly demonstrate the existence of speech information in EEG signals and lay the foundational stone for future research in speech assistive technologies.

Recent modeling of brain activities encompasses the fusion of different modalities. However, fusing brain modalities requires not only the efficient and compatible representation of the signals but also the benefits associated with it. For instance, the combination of the functional characteristics of EEGs with the structural features of functional magnetic resonance imaging contributes to a better interpretation localization of brain activities. In this paper, we consider the EEG signals as parallel 2D string images from which we extract their visual abstract representations of EEG features. This representation can benefit not only the EEG modeling of the signals but also a future fusion with another modality, like fMRI. In particular, the new methodology, called Bar-LG, provides a reduced discretization of the EEG signals into selected minima/maxima in order to be used in a form of tokens for EEG brain activities of interest. A formal context-free language is used to express and represent the extracted tokens for the selected active brain regions. Then, a Generalized Stochastic Petri-Nets (GSPN) model is used for expressing the functional associations and interactions of these EEG signals as 2D image regions. An illustrative EEG example of epileptic seizure is presented to show the Bar-LG methodology's abstract capabilities.

During sleep., breathing-related sleep disorders (BSD) are very probable to cause distortions on human health and even be life-threatening. Among the different types of BSD., apnea accounts for one of the most common. Many detection algorithms have been proposed for spotting and classifying apneas, using one feature or being designed for binary classification. Also, many proposed clinical setups for respiratory data acquisition are invasive, making the application to patients a non-trial task. In this study, we aim to propose an easy-to-apply and patient-friendly clinical setup with a BSD detection that utilizes a multi-feature classification scheme for binary (apnea, healthy), as well as multiple classes (healthy, central, mixed, and obstructive apneas and hypopneas). Our clinical setup includes a high-resolution microphone attached to the bed at a very close distance to the patient. Our multi-feature approach contains spectral, statistical, and symbolic-based characteristics of respiratory signals of five patients admitted for a first BSD diagnosis and assesses the performance of different classification algorithms iteratively. The results show a high classification performance ($>$98% for binary and$>$84% for multi-class classification) for either classification scheme. A robust classification scheme is thus proposed, utilizing the entire content of the recorded respiratory signal. Such a classification scheme leads to a promising result towards the design of portable devices with multi-features for real-time detection of BSD.

Electrocardiogram (ECG) signal constitutes a valuable technique that provides considerable information towards the early diagnosis of several cardiovascular diseases, especially regarding the detection of abnormal heart rate, namely arrhythmias. In this paper, innovative methodologies that allow for the efficient classification of cardiac rhythm are presented. The proposed methods are based on ECG signal analysis, extraction of significant features, as well as classification algorithms. Several clinical, time- and frequency-domain features are either calculated, or automatically extracted by means of a Convolutional Neural Network, while traditional machine learning algorithms, such as k-Nearest Neighbors and Random Forests are employed in order to classify the ECG signals among 7 different cases of abnormal and normal heart rate. The learning methods are carried out within the JADBio software tool, that also performs feature selection prior to classification. The experimental results demonstrate high performance of the deployed methods in terms of relevant statistical metrics, while they yielded an average validation Area Under the Curve (AUC) of 99.9%.

Epilepsy is one of the most common brain disorders worldwide. The basic principle in epilepsy is to resect the epileptogenic zone (EZ) when the medicaments are inadequate to suppress epileptic seizures. Epilepsy is accompanied by interictal spikes, a surrogate marker serving as an identifier of seizures. The automatic temporal detection of these spikes is of major importance due to the demanding time consumption of the manual annotation. Electro- and magneto- encephalography (EEG and MEG) are the most usual measurement modalities for the recording of brain activity. EEG and MEG are ideal modalities for the non-invasive monitoring of drug-resistant epilepsy. Many approaches have been proposed for the temporal detection of interictal spikes. However, only single measurement modality (EEG or MEG) has been used up to now, neglecting their complementary content. In this study, we develop a multi-feature and iterative classification scheme with input from either single modality (EEG or MEG) or combined EEG/MEG (EMEG). The inputs include statistical (kurtosis and Renyi Entropy) and spectral (Energy) features as well as the functional connectivity metrics, global and local efficiency from imaginary phase lag index networks. The classification performance for all modalities ranges from 89% to 92.8%, with the maximum performance being observed for EMEG. Overall, the complementarity of EEG and MEG on the detection of interictal spikes is promising, opening new considerations on the development of automatic epileptic spike detection approaches.

Aquaculture faces the issue of net integrity on cage farming. Holes on the net need to be detected but as yet the process is not fully automated. This work is a second-generation embedded system to detect in real time holes in aquaculture nets from a video input. It extends previous results by processing video rather than still images, under lighting variation, haze, and different size of holes along each frame. The modeling and simulation of the new algorithm has been done in MATLAB; the system has been designed and implemented on a Field Programmable Gate Array (FPGA) - based platform. The proposed system has substantially better performance vs. software at a much lower energy consumption.

Agriculture is by its nature a complicated scientific field, related to a wide range of expertise, skills, methods and processes which can be effectively supported by computerized systems. There have been many efforts towards the establishment of an automated agriculture framework, capable to control both the incoming data and the corresponding processes. The recent advances in the Information and Communication Technologies (ICT) domain have the capability to collect, process and analyze data from different sources while materializing the concept of agriculture intelligence. The thriving environment for the implementation of different agriculture systems is justified by a series of technologies that offer the prospect of improving agricultural productivity through the intensive use of data. The concept of big data in agriculture is not exclusively related to big volume, but also on the variety and velocity of the collected data. Big data is a key concept for the future development of agriculture as it offers unprecedented capabilities and it enables various tools and services capable to change its current status. This survey paper covers the state-of-the-art agriculture systems and big data architectures both in research and commercial status in an effort to bridge the knowledge gap between agriculture systems and exploitation of big data. The first part of the paper is devoted to the exploration of the existing agriculture systems, providing the necessary background information for their evolution until they have reached the current status, able to support different platforms and handle multiple sources of information. The second part of the survey is focused on the exploitation of multiple sources of information, providing information for both the nature of the data and the combination of different sources of data in order to explore the full potential of ICT systems in agriculture.

Fish‐cage dysfunction in aquaculture installations can trigger significant negative consequences affecting the operational costs. Low oxygen levels, due to excessive fooling's, leads to decrease growth performance, and feed efficiency. Therefore, frequent periodic inspection of fish‐cage nets is required, but this task can become quite expensive with the traditional means of employing professional divers that perform visual inspections at regular time intervals. The modern trend in aquaculture is to take advantage of IT technologies with the use of a small‐sized, low‐cost autonomous underwater vehicle, permanently residing within a fish cage and performing regular video inspection of the infrastructure for the entire net surface. In this study, we explore specialised image processing schemes to detect net holes of multiple area size and shape. These techniques are designed with the vision to provide robust solutions that take advantage of either global or local image structures to provide the efficient inspection of multiple net holes.

This paper reviews state-of-the-art research solutions across the spectrum of medical imaging informatics, discusses clinical translation, and provides future directions for advancing clinical practice. More specifically, it summarizes advances in medical imaging acquisition technologies for different modalities, highlighting the necessity for efficient medical data management strategies in the context of AI in big healthcare data analytics. It then provides a synopsis of contemporary and emerging algorithmic methods for disease classification and organ/ tissue segmentation, focusing on AI and deep learning architectures that have already become the de facto approach. The clinical benefits of in-silico modelling advances linked with evolving 3D reconstruction and visualization applications are further documented. Concluding, integrative analytics approaches driven by associate research branches highlighted in this study promise to revolutionize imaging informatics as known today across the healthcare continuum for both radiology and digital pathology applications. The latter, is projected to enable informed, more accurate diagnosis, timely prognosis, and effective treatment planning, underpinning precision medicine.

In recent years, several approaches have been introduced for estimating the spike onset zone within the irritative zone in epilepsy diagnosis for presurgical planning. One important direction utilizes source analysis from combined electroencephalography (EEG) and magnetoencephalography (MEG), EMEG, leveraging the benefits from the complementary properties of the two modalities. For EMEG source reconstruction, an average across the annotated epileptic spikes is often used to improve the signal-to-noise-ratio (SNR). In this contribution, we propose a two-phase clustering of interictal spikes with unsupervised learning methods, namely Self Organizing Maps (SOM) and K-means. In addition, we investigate the accuracy of combined EMEG source analysis on the sorted activity, using an individualized (with regard to both geometry and conductivity) six-compartment finite element head model with calibrated skull conductivity and white matter conductivity anisotropy. The results indicate that SOM eliminates the random variations of K-means and stabilizes the clustering efficiency. In terms of source reconstruction accuracy, this study demonstrates that the combined use of modalities reveals activity around two focal cortical dysplasias (FCDs), of one epilepsy patient, one in the right frontal area and one smaller in the left premotor cortex. It is worth mentioning that only EMEG could localize the left premotor FCD, which was then also found in surgery to be the responsible for triggering the epilepsy.

This paper evaluates the usage of matching pursuit (MP) features of electroencephalographic (EEG) signals and classification techniques on automatic absence seizure detection. Absence epileptic seizures are neurological disorders which are manifested as abnormal EEG patterns. Matching pursuit algorithm is able to decompose a signal into components with specific time-frequency characteristics. It is a robust technique especially when there is complex, multicomponent signal. In the present study, a clinical dataset containing 40 annotated absence seizures in long-term EEG recordings from pediatric epileptic patients (with age 6.0±2.9 years) was analyzed. The extracted MP features fed an automatic classification schema which achieved a time window based discrimination accuracy of 98.5%. As indicated by the study's results, the proposed features and analysis methods can be a promising addition to the area of automatic absence seizures detection.

This paper investigates the effectiveness of Common Spatial Patterns (CSP) analysis of EEG signals on the automatic detection of focal epileptic seizures. Focal seizures are characterized by unilaterally triggered abnormal brain activity. CSP analysis has been frequently used in literature for multichannel EEG signal separation between two states. In the present study, EEG recordings from 10 subjects aged 7.7±4.4 years, including 63 seizures, were analyzed with respect to seizure detection and discrimination between interictal and ictal periods. Machine learning techniques of feature selection and classification were used in the analysis, resulting in a best achieved classification accuracy of 91.1%.

Coxibs are a group of drugs with selective inhibition against cyclooxygenase-2 (COX-2) enzymes with increased interest from scientific community due to their side effects and potential other pharmacological mechanisms. The aim of this work is to utilize the chemical characteristics of coxibs in order to identify compounds with similar chemical structure. The approach is based on the assessment of the Simplified Molecular-Input Line-Entry System (SMILES) as adequate molecular structure representations for the identification of drug similarities. The similarity measurements are based on molecular fingerprints that were extracted from coxibs and the Maximum Consecutive Subsequence (MCS) algorithm. An ensemble of methods based on majority voting, weighting and equal weighting on the algorithms was further applied. Majority voting returned 200 similar compounds whereas weighting and equal weighting returned 53 and 27 compounds respectively. Interestingly, despite the independence of the methods, all three identified 20 common compounds. The identification of drugs with potential chemical similarity with coxibs, as revealed from similarity measurements of fingerprints and MCS scores could provide new insights for potential biological targets for coxibs or drugs that could interact with COX-2 or other biological targets of coxibs.

Many studies have addressed the relations between different human brain regions and their role in cognitive, motor and sensory functions in patients that have suffered a brain lesion (stroke, traumatic brain injury, tissue removal). Nowadays, it is well established that the brain works as a network and the symptoms in a person are a combination of the direct impact of the lesion in a single region and its connectivity with other healthy brain regions. The aim of the present study is the development of a user-friendly desktop application to predict the induced cognitive deficits in patients who have suffered a brain lesion. The herein presented application is based on Neurosynth platform, and takes as an input a MRI mask that describes a lesion. Then our software exploits the knowledge that already exists in Neurosynth platform, so as to predict the potential deficits by grouping the Neurosynth's terms that have increased Z scores with our mask. In addition, we have embedded two types of visualization methods: One to present the slices of the brain mask and another to show the 3D volume of the mask into 3D semitransparent human brain. The added value of the presented application is that it may give us a clue about which mechanisms are probably affected by a lesion in a specific region, while in the future it could provide neurosurgeons with insightful knowledge helping them in the plannification of a forthcoming surgical procedure. The proposed software was tested on 7 stroke patients, predicting accurately the 91% of the measured deficits found during a neuropsychological assessment.

The primary somatosensory cortex remains one of the most investigated brain areas. However, there is still an absence of an integrated methodology to describe the early temporal alterations in the primary somatosensory network. Source analysis based on combined Electro-(EEG) and Magneto-(MEG) Encephalography (EMEG) has been recently shown to outperform the one's based on single modality EEG or MEG. The study and potential of combined EMEG form the goal of the current study, which investigates the time-variant connectivity of the primary somatosensory network. A subject-individualized pipeline combines a functional source separation approach with the effective connectivity analysis of different spatiotemporal source patterns using a realistic and skull-conductivity calibrated head model. Three-time windows are chosen for each modality EEG, MEG, and EMEG to highlight the thalamocortical and corticocortical interactions. The results show that EMEG is promising in suppressing a so-called connectivity 'leakage' effect when later components seem to influence earlier components, just due to too similar leadfields. Our current results support the notion that EMEG is superior in suppressing the spurious flows within a network of very rapid alterations.

The human cerebellum contains almost 50% of the neurons in the brain, although its volume does not exceed 10% of the total brain volume. The goal of this study is to derive the functional network of the cerebellum during the resting-state and then compare the ensuing group networks between males and females. Toward this direction, a spatially constrained version of the classic spectral clustering algorithm is proposed and then compared against conventional spectral graph theory approaches, such as spectral clustering, and N-cut, on synthetic data as well as on resting-state fMRI data obtained from the Human Connectome Project (HCP). The extracted atlas was combined with the anatomical atlas of the cerebellum resulting in a functional atlas with 46 regions of interest. As a final step, a gender-based network analysis of the cerebellum was performed using the data-driven atlas along with the concept of the minimum spanning trees. The simulation analysis results confirm the dominance of the spatially constrained spectral clustering approach in discriminating activation patterns under noisy conditions. The network analysis results reveal statistically significant differences in the optimal tree organization between males and females. In addition, the dominance of the left VI lobule in both genders supports the results reported in a previous study of ours. To our knowledge, the extracted atlas comprises the first resting-state atlas of the cerebellum based on HCP data.

The application of machine learning methods for the identification of candidate genes responsible for phenotypes of interest, such as cancer, is a major challenge in the field of bioinformatics. These lists of genes are often called genomic signatures and their linkage to phenotype associations may form a significant step in discovering the causation between genotypes and phenotypes. Traditional methods that produce genomic signatures from DNA Microarray data tend to extract significantly different lists under relatively small variations of the training data. That instability hinders the validity of research findings and raises skepticism about the reliability of such methods. In this study, a complete framework for the extraction of stable and reliable lists of candidate genes is presented. The proposed methodology enforces stability of results at the validation step and as a result, it is independent of the feature selection and classification methods used. Furthermore, two different statistical tests are performed in order to assess the statistical significance of the observed results. Moreover, the consistency of the signatures extracted by independent executions of the proposed method is also evaluated. The results of this study highlight the importance of stability issues in genomic signatures, beyond their prediction capabilities.

The IEEE International Conference on Biomedical and Health Informatics (BHI) is a special topic conference of the IEEE Engineering in Medicine and Biology Society (IEEEEMBS). BHI2017 was co-located with the annual HIMSS Conference & Exhibition in Rosen Plaza Hotel, Orlando, Florida, USA during Feb. 16-19, 2017. The focus of BHI2017 was on "informatics for personalized, precision and preventive healthcare." Advancing health informatics has been identified as a grand challenge for engineering in the 21st century by the National Academy of Engineering. Managing and improving human health will require novel and creative informatics solutions to accelerate scientific discovery, translate innovations into successful treatments, re-engineer care practices and infrastructure, all with effective use of hardware and software. BHI2017 provided a unique forum showcasing enabling technologies in both methodology and clinical advancement in data/information acquisition, transmission, storage, retrieval, visualization, processing, analysis, interpretation and validation. In addition, it demonstrated how integrative informatics solutions can be employed in novel clinical applications, and how the deployment of integrated bioinformatics, with the support of Internet of Things (IoT), can facilitate precision and preventive medicine. In this special issue, six papers are selected to exhibit recent development in areas highlighted in BHI2017. The papers are briefly summarized here.

The discrimination of the preictal state in EEG signals is of great importance in neuroscience and the epileptic seizure prediction field has yet to provide conclusive evidence. In this study, three different classification approaches, including the Repeated Incremental Pruning to Produce Error Reduction (RIPPER) algorithm, Support Vector Machines (SVMs) and Neural Networks (NNs), are investigated for their ability to discriminate preictal from interictal EEG segments. Using public EEG data, a wide range of features is extracted from each segment and then applied to the classifiers. The analysis covers a patient-specific approach, so as to optimize the decision to each patient individually and a patient-independent approach in order to explore a global prediction approach that can discriminate randomly selected preictal and interictal segments from all patients. Overall, the first approach aims at revealing patient-specific epileptic characteristics, whereas the second seeks for potential general preictal-related signs. The results reveal that in the patient-specific case, the SVM classifier exhibits the highest classification accuracy in both preictal and interictal classes reaching 85.75% sensitivity and specificity. As it is expected, the classification performance is lower for the patient-independent case at 68.5%, due to the complicated nature of preictal activity and the variations among patients condition.

We present a novel synergistic methodology for the spatio-temporal analysis of single Electroencephalogram (EEG) trials. This new methodology is based on the novel synergy of Local Global Graph (LG graph) to characterize define the structural features of the EEG topography as a global descriptor for robust comparison of dominant topographies (microstates) and Hidden Markov Models (HMM) to model the topographic sequence in a unique way. In particular, the LG graph descriptor defines similarity and distance measures that can be successfully used for the difficult comparison of the extracted LG graphs in the presence of noise. In addition, hidden states represent periods of stationary distribution of topographies that constitute the equivalent of the microstates in the model. The transitions between the different microstates and the formed syntactic patterns can reveal differences in the processing of the input stimulus between different pathologies. We train the HMM model to learn the transitions between the different microstates and express the syntactic patterns that appear in the single trials in a compact and efficient way. We applied this methodology in single trials consisting of normal subjects and patients with Progressive Mild Cognitive Impairment (PMCI) to discriminate these two groups. The classification results show that this approach is capable to efficiently discriminate between control and Progressive MCI single trials. Results indicate that HMMs provide physiologically meaningful results that can be used in the syntactic analysis of Event Related Potentials.

Pathway analysis methodologies couple traditional gene expression analysis with knowledge encoded in established molecular pathway networks, offering a promising approach towards the biological interpretation of phenotype differentiating genes. Early pathway analysis methodologies, named as gene set analysis (GSA), view pathways just as plain lists of genes without taking into account either the underlying pathway network topology or the involved gene regulatory relations. These approaches, even if they achieve computational efficiency and simplicity, consider pathways that involve the same genes as equivalent in terms of their gene enrichment characteristics. Most recent pathway analysis approaches take into account the underlying gene regulatory relations by examining their consistency with gene expression profiles and computing a score for each profile. Even with this approach, assessing and scoring single-relations limits the ability to reveal key gene regulation mechanisms hidden in longer pathway sub-paths. We introduce MinePath, a pathway analysis methodology that addresses and overcomes the aforementioned problems. MinePath facilitates the decomposition of pathways into their constituent sub-paths. Decomposition leads to the transformation of single-relations to complex regulation sub-paths. Regulation sub-paths are then matched with gene expression sample profiles in order to evaluate their functional status and to assess phenotype differential power. Assessment of differential power supports the identification of the most discriminant profiles. In addition, MinePath assess the significance of the pathways as a whole, ranking them by their p-values. Comparison results with state-of-the-art pathway analysis systems are indicative for the soundness and reliability of the MinePath approach. In contrast with many pathway analysis tools, MinePath is a web-based system (www.minepath.org) offering dynamic and rich pathway visualization functionality, with the unique characteristic to color regulatory relations between genes and reveal their phenotype inclination. This unique characteristic makes MinePath a valuable tool for in silico molecular biology experimentation as it serves the biomedical researchers' exploratory needs to reveal and interpret the regulatory mechanisms that underlie and putatively govern the expression of target phenotypes.

Obtaining an up-to-date high-resolution description of land cover is a challenging task due to the high cost and labor-intensive process of human annotation through field studies. This work introduces a radically novel approach for achieving this goal by exploiting the proliferation of remote sensing satellite imagery, allowing for the up-to-date generation of global-scale land cover maps. We propose the application of multilabel classification, a powerful framework in machine learning, for inferring the complex relationships between the acquired satellite images and the spectral profiles of different types of surface materials. Introducing a drastically different approach compared to unsupervised spectral unmixing, we employ contemporary ground-collected data from the European Environment Agency to generate the label set and multispectral images from the MODIS sensor to generate the spectral features, under a supervised classification framework. To validate the merits of our approach, we present results using several state-of-the-art multilabel learning classifiers and evaluate their predictive performance with respect to the number of annotated training examples, as well as their capability to exploit examples from neighboring regions or different time instances. We also demonstrate the application of our method on hyperspectral data from the Hyperion sensor for the urban land cover estimation of New York City. Experimental results suggest that the proposed framework can achieve excellent prediction accuracy, even from a limited number of diverse training examples, surpassing state-of-the-art spectral unmixing methods.

The importance of the development of the Primary Care network in a developed country is indisputable high. The increasing pressure for productivity improvement and reduction of costs, requires activities focusing on the control and optimization of care processes improving their efficiency and effectivity. A keystone in such priority is the utilization of several effective and comprehensive processes in the everyday practice. The aim of this study is the development of a framework for the improvement of the Health Care processes introducing Business Process Modeling Notations (BPMNs) methodologies. The large majority of processes in the organizations and healthcare centers which belong to the Greek Primary Care Healthcare system have been collected. For each process the BPMNs diagram has been developed and presented. The analysis and the quantitative indexes have been captured by the integration of the processes in a web application that has been designed and implemented. The application's outcome has been analyzed partially since it will be tested in a Health Care Center in Kissamos, Crete, Greece.

Cancer-tumor growth is a complex process depending on several biological factors, such as the chemical microenvironment of the tumor, the cellular metabolic profile, and its proliferation rate. Several mathematical models have been developed for identifying the interactions between tumor cells and tissue microenvironment, since they play an important role in tumor formation and progression. Toward this direction we propose a new continuum model of avascular glioma-tumor growth, which incorporates a new factor, namely, the glycolytic potential of cancer cells, to express the interactions of three different tumor-cell populations (proliferative, hypoxic, and necrotic) with their tissue microenvironment. The glycolytic potential engages three vital nutrients, i.e., oxygen, glucose, and lactate, which provide cells with the necessary energy for their survival and proliferation. Extensive simulations are performed for different evolution times and various proliferation rates, in order to investigate how the tumor growth is affected. According to medical experts, the experimental observations indicate that the model predicts quite satisfactorily the overall tumor growth as well as the expansion of each region separately. Following extensive evaluation, the proposed model may provide an essential tool for patient-specific tumor simulation and reliable prediction of glioma spatiotemporal expansion.

Wilms tumor, or nephroblastoma, is a cancer of the kidneys that typically occurs in children and rarely in adults. Around 10% of Wilms tumor patients are diagnosed having a concurrent syndrome that enhances the risk of Wilms tumor. A screening method for early detection of Wilms tumor in these patients would be beneficial, since the size or stage of a tumor is related to outcome. We introduce a miRNA pathway analysis methodology that takes into account the topology and regulation mechanisms of the gene regulatory networks and identify disrupted sub-paths in known pathways, using miRNA- expressions. The methodology was applied on a miRNA-expression study and a predictive model was developed, using machine-learning (decision-tree induction) approaches. The model is able to identify putative mechanisms that underlie and govern the Wilms tumor phenotype, and discriminate between diseased and healthy subjects. Initial experimental results are promising and in line with the relevant biomedical literature.

Human brain is the most complicated network and its functional mechanism is a demanding concept in neuroscience research. Graph theory and forms an interesting tool for modeling the brain interactions and estimated brain parameters. In this paper, we consider synchronization features for modeling brain operations in electro-encephalogram (EEG) responses to kanizsa and fractal stimuli, using minimal spanning tree (MST) on a network of phase synchronization EEG channels. Graphs of phase-synchronization activity and MST structures are computed using these graphs. The proposed approach yields evidence that the fractal stimuli generate stronger energy response and synchronization of theta band in occipital lobe.

The diagnosis and treatment of brain gliomas has still many weaknesses and is followed by a high mortality rate and very low life expectancy. Facing towards the tendency for a personalized diagnosis and therapy, any effort should be focused to the spatiotemporal growth of the tumor for the early prognosis and treatment of the gliomas. In the above context, this study aims to analyze the brain markers choline (Cho) and fractional anisotropy (FA) in order to determine whether they are reliable indices of glioma presence in a brain region, before this region is damaged to such an extent that will be visible in the classic magnetic resonance imaging. This study is based on the experimental analysis of Cho from the spectrum of metabolites and FA from the diffusion tensor image, aiming ultimately to the improvement of the diagnosis and thus the therapy and the life expectancy rate of the patients suffering from glioma.

Biological networks in living organisms can be seen as the ultimate means of understanding the underlying mechanisms in complex diseases, such as oral cancer. During the last decade, many algorithms based on high-throughput genomic data have been developed to unravel the complexity of gene network construction and their progression in time. However, the small size of samples compared to the number of observed genes makes the inference of the network structure quite challenging. In this study, we propose a framework for constructing and analyzing gene networks from sparse experimental temporal data and investigate its potential in oral cancer. We use two network models based on partial correlations and kernel density estimation, in order to capture the genetic interactions. Using this network construction framework on real clinical data of the tissue and blood at different time stages, we identified common disease-related structures that may decipher the association between disease state and biological processes in oral cancer. Our study emphasizes an altered MET (hepatocyte growth factor receptor) network during oral cancer progression. In addition, we demonstrate that the functional changes of gene interactions during oral cancer progression might be particularly useful for patient categorization at the time of diagnosis and/or at follow-up periods.

Breast cancer is a highly heterogeneous disease and very common among western women. The main cause of death is not the primary tumor but its metastases at distant sites, such as lymph nodes and other organs (preferentially lung, liver, and bones). The study of circulating tumor cells (CTCs) in peripheral blood resulting from tumor cell invasion and intravascular filtration highlights their crucial role concerning tumor aggressiveness and metastasis. Genomic research regarding CTCs monitoring for breast cancer is limited due to the lack of indicative genes for their detection and isolation. Instead of direct CTC detection, in our study, we focus on the identification of factors in peripheral blood that can indirectly reveal the presence of such cells. Using selected publicly available breast cancer and peripheral blood microarray datasets, we follow a two-step elimination procedure for the identification of several discriminant factors. Our procedure facilitates the identification of major genes involved in breast cancer pathology, which are also indicative of CTCs presence.

Clustering analysis based on temporal profile of genes may provide new insights in particular biological processes or conditions. We report such an integrative clustering analysis which is based on the expression patterns but is also influenced by temporal changes. The proposed platform is illustrated with a temporal gene expression dataset comprised of pellet culture-conditioned human primary chondrocytes and human bone marrow-derived mesenchymal stem cells (MSCs). We derived three clusters in each cell type and compared the content of these classes in terms of temporal changes. We further considered the induced biological processes and the gene-interaction networks formed within each cluster and discuss their biological significance. Our proposed methodology provides a consistent tool that facilitates both the statistical and biological validation of temporal profiles through spatial gene network profiles.

The understanding of the mechanisms of human brain is a demanding issue for neuroscience research. Physiological studies acknowledge the usefulness of synchronization coupling in the study of dysfunctions associated with reading difficulties. Magnetoencephalogram (MEG) is a useful tool towards this direction having been assessed for its superior accuracy over other modalities. In this paper we consider synchronization features for identifying brain operations. Independent Component Analysis (ICA) is applied on MEG surface signals in controls and children with reading difficulties and are clustered to representative components. Then, coupling measures of mutual information and partial directed coherence are estimated in order to reveal dysfunction of cerebral networks and its related coordination.

In the recent years microarray technologies have gained a lot of popularity for their ability to quickly measure the expression of thousands of genes and provide valuable information for linking complex diseases such as cancer to their genetic underpinnings. Nevertheless the large number of parameters to be estimated in relation to the small number of available samples gives rise to an “ill posed” problem where the possible solution is not stable under slight changes either in the dataset or the initial conditions and starting points. In this work we present a generic classification framework that works in an iterative manner and converges to a stable solution that combines good accuracy with biologically meaningful feature selection. The methodology is orthogonal to the specific classification algorithm used. We compare some of the most widely used classifiers based on their average discrimination power and the size of the derived gene signature. According to our proposed model named Stable Bootstrap Validation (SBV), a unified `77 common-gene signature' was selected, which is closely associated with several aspects of breast tumorigenesis and progression, as well as patient-specific molecular and clinical characteristics.

The importance of ErbB receptor signaling in breast cancer is consistent with its functional role in normal development of mammary gland. The study of the ErbB signaling network and its bidirectional cross-talk with hormonal receptors, such as estrogen receptor (ER) encloses information about the molecular mechanisms on breast cancer evolution, progression and endocrine resistance. With this analysis we attempt to examine the differences in activation/inhibition of intracellular signaling molecules within ErbB signaling cascade on ER+ and ER-breast cancer patients. With the proposed framework we model the genetic interactions in the ErbB signaling pathway directly from expression data as Gaussian approximations and compare them with the KEGG canonical ErbB pathway in order to identify significant molecular deformations characterizing the studied population. The results indicate a distinct profile of activation/inhibition between the two ER populations and highlight the primary role of PI3K/Akt pathway in breast cancer progression and targeted treatment strategies.

Stents are medical devices used in cardiovascular intervention for unblocking the diseased arteries and restoring blood flow. During stent implantation the deformation of the arterial wall as well as the resulted stresses caused in the arterial morphology are studied. In this paper we study the effect of the composition of the atherosclerotic plaque during the stent deployment procedure, using Finite Element modeling. The stenting procedure is simulated for two different cases; in the first the presence of the plaque is ignored whereas in the second a three dimensional (3D) stiff calcified plaque is located in the stenotic area of the artery. Results indicate that in the second case the von Mises stresses in the arterial wall are higher than the stresses occurred in the first case. In addition, the distribution of the arterial von Mises stress depends on the plaque composition.

Cancer growth is a complex process which depends on several tasks that cancer cells have to perform in order to live and proliferate. Among these tasks, perhaps the most significant one is to reach adequate sources of nutrients, such as oxygen and glucose, from their surrounding environment in order to initiate their respiration process and provide them with the necessary energy in the form of ATP molecules. Cellular respiration is a biological mechanism that consists of two sequential processes, named `glycolysis' and `oxidative phosporylation' (OXPHOS). Since 1956, when the biologist Otto Warburg discovered the increase of glycolysis in cancer cell compared to healthy cells, glycolysis has been studied in depth in order to understand its role in cancer genesis and growth. Towards this direction we propose a new in-silico cancer growth model which embeds the glycolytic potential of the cancer cells in the growth process. The experimental observations obtained show that the model fits the cancer data predicting the tumor's growth, in the proliferative, hypoxic and necrotic zone, quite satisfactory.

The proposed analysis considers aspects of both statistical and biological validation of the glycolysis effect on brain gliomas, at both genomic and metabolic level. In particular, two independent datasets are analyzed in parallel, one engaging genomic (Microarray Expression) data and the other metabolomic (Magnetic Resonance Spectroscopy Imaging) data. The aim of this study is twofold. First to show that, apart from the already studied genes (markers), other genes such as those involved in the human cell glycolysis significantly contribute in gliomas discrimination. Second, to demonstrate how the glycolysis process can open new ways towards the design of patient-specific therapeutic protocols. The results of our analysis demonstrate that the combination of genes participating in the glycolytic process (ALDOA, ALDOC, ENO2, GAPDH, HK2, LDHA, LDHB, MDH1, PDHB, PFKM, PGI, PGK1, PGM1 and PKLR) with the already known tumor suppressors (PTEN, Rb, TP53), oncogenes (CDK4, EGFR, PDGF) and HIF-1, enhance the discrimination of low versus high-grade gliomas providing high prediction ability in a cross-validated framework. Following these results and supported by the biological effect of glycolytic genes on cancer cells, we address the study of glycolysis for the development of new treatment protocols.

Biological networks are often described as probabilistic graphs in the context of gene and protein sequence analysis in molecular biology. Microarrays and proteomics technology allow the monitoring of expression levels over thousands of biological units over time. In experimental efforts we are interested in unveiling pairwise interactions. Many graphical models have been introduced in order to discover associations from the expression data analysis. However, the small size of samples compared to the number of observed genes/proteins makes the inference of the network structure quite challenging. In this study we generate gene-protein networks from sparse experimental data using two methods, partial correlations and Kernel Density Estimation, in order to capture genetic interactions. Dynamic Gaussian analysis is used to match special characteristics to genes and proteins at different time stages utilizing the KDE method for expressing Gaussian associations with non-linear parameters.

It is generally accepted that evoked and induced activations represent different aspects of cerebral functions during an Event Related Potentials (ERP) experiment. Independent Component Analysis (ICA) has been successfully applied to event related electroencephalography (EEG) to decompose it into a sum of spatially fixed and temporally independent components that can be attributed to underlying cortical activity. A major problem in the application of ICA is the stability of estimated independent components. In this paper we exploited the split-half approach to assess component stability. We used different measures quantifying both phase and energy aspects of the ERP, in order to distinguish evoked from induced oscillations. We applied these measures to the stable independent components derived from a dataset of progressive Mild Cognitive Impairment (PMCI) and elderly controls. We found reduced energy in the induced theta activity in PMCI subjects, in accordance with previous studies. In addition, PMCI subjects presented lower phase-locking values and diminished late alpha band energy in contrast to controls.

The biological processes are widely studied by genome analysis leading to a large number of genes, thus making necessary the use of automated evaluation methods. In this study, we examine the influence of algorithmic parameters in the prediction power of a gene signature and in the selection process of the signature itself. We focus on one gene selection approach applied on a dataset of the budding yeast Saccharomyces cerevisiae, using quite different parameters and evaluate the influence on the selected signature. In particular, we adopt a recursive feature elimination process where at each step the prognostic power of the set of remaining genes is evaluated by five different classifiers, as well as by four classifier-fusions schemes. More specifically, we consider the logistic-sigmoid, kernel nearest centroid, kernel minimum squared error, kernel subspace, and support vector machines as classifiers with different parameters and/or kernel functions. We also study four fusion methods in order to reduce uncertainties related to the classifier evaluating the prognostic significance of genes. In all cases, the selection process is embedded into a cross validation scheme in order to enhance the confidence on the generalization of results. We consider the differences of signatures based on gene overlap and also the biological annotation of selected genes, using the MIPS FunCat architecture. We found out that a robust identification of a number of highly differential genes can offer “good” predictive power to the models. Furthermore, the classification accuracy achieved by mixtures of experts can be significantly better than the one of the individual classifiers. We also pointed out that different selection schemes result in a diverse size of gene signature, with differences in the selected genes. Nevertheless, when we annotate the genes of each signature we find that the same biological processes are invoked, with possibly small differences in the relative frequency of participation.

Glioma, especially glioblastoma, is a leading cause of brain cancer fatality involving highly invasive and neoplastic growth. Diffusive models of glioma growth use variations of the diffusion-reaction equation in order to simulate the invasive patterns of glioma cells by approximating the spatiotemporal change of glioma cell concentration. The most advanced diffusive models take into consideration the heterogeneous velocity of glioma in gray and white matter, by using two different discrete diffusion coefficients in these areas. Moreover, by using diffusion tensor imaging (DTI), they simulate the anisotropic migration of glioma cells, which is facilitated along white fibers, assuming diffusion tensors with different diffusion coefficients along each candidate direction of growth. Our study extends this concept by fully exploiting the proportions of white and gray matter extracted by normal brain atlases, rather than discretizing diffusion coefficients. Moreover, the proportions of white and gray matter, as well as the diffusion tensors, are extracted by the respective atlases; thus, no DTI processing is needed. Finally, we applied this novel glioma growth model on real data and the results indicate that prognostication rates can be improved.

Glioma is one of the most aggressive types of brain tumor. Several mathematical models have been developed during the past two decades, toward simulating the mechanisms that govern the development of glioma. The most common models use the diffusion-reaction equation (DRE) for simulating the spatiotemporal variation of tumor cell concentration. Nevertheless, despite the applications presented, there has been little work on studying the details of the mathematical solution and implementation of the 3-D diffusion model and presenting a qualitative analysis of the algorithmic results. This paper presents a complete mathematical framework on the solution of the DRE using different numerical schemes. This framework takes into account all characteristics of the latest models, such as brain tissue heterogeneity, anisotropic tumor cell migration, chemotherapy, and resection modeling. The different numerical schemes presented have been evaluated based upon the degree to which the DRE exact solution is approximated. Experiments have been conducted both on real datasets and a test case for which there is a known algebraic expression of the solution. Thus, it is possible to calculate the accuracy of the different models.

The concept of gene signature overlap has been addressed previously in a number of research papers. A common conclusion is the absence of significant overlap. In this paper, we verify the aforementioned fact, but we also assess the issue of similarities not on the gene level, but on the biology level hidden underneath a given signature. We proceed by taking into account the biological knowledge that exists among different signatures, and use it as a means of integrating them and refining their statistical significance on the datasets. In this form, by integrating biological knowledge with information stemming from data distributions, we derive a unified signature that is significantly improved over its predecessors in terms of performance and robustness. Our motive behind this approach is to assess the problem of evaluating different signatures not in a competitive but rather in a complementary manner, where one is treated as a pool of knowledge contributing to a global and unified solution.

Although magnetic resonance spectroscopy (MRS) methods of 1.5Tesla (T) and 3T have been widely applied during the last decade for noninvasive diagnostic purposes, only a few studies have been reported on the value of the information extracted in brain cancer discrimination. The purpose of this study is threefold. First, to show that the diagnostic value of the information extracted from two different MRS scanners of 1.5T and 3T is significantly influenced in terms of brain gliomas discrimination. Second, to statistically evaluate the discriminative potential of publicly known metabolic ratio markers, obtained from these two types of scanners in classifying low-, intermediate-, and high-grade gliomas. Finally, to examine the diagnostic value of new metabolic ratios in the discrimination of complex glioma cases where the diagnosis is both challenging and critical. Our analysis has shown that although the information extracted from 3T MRS scanner is expected to provide better brain gliomas discrimination; some factors like the features selected, the pulse-sequence parameters, and the spectroscopic data acquisition methods can influence the discrimination efficiency. Finally, it is shown that apart from the bibliographical known, new metabolic ratio features such as N-acetyl aspartate/ S, Choline/ S, Creatine/ S , and myo-Inositol/ S play significant role in gliomas grade discrimination.

An increasing number of studies have profiled gene expressions in tumor specimens using distinct microarray platforms and analysis techniques. One challenging task is to develop robust statistical models in order to integrate multi-platform findings. We compare some methodologies on the field with respect to estrogen receptor (ER) status, and focus on a unified-among-platforms scale implemented by Shen et al. in 2004, which is based on a Bayesian mixture model. Under this scale, we study the ER intensity similarities between four breast cancer datasets derived from various platforms. We evaluate our results with an independent dataset in terms of ER sample classification, given the derived gene ER signatures of the integrated data. We found that integrated multi-platform gene signatures and fold-change variability similarities between different platform measurements can assist the statistical analysis of independent microarray datasets in terms of ER classification.

BACKGROUND: Information extraction from microarrays has not yet been widely used in diagnostic or prognostic decision-support systems, due to the diversity of results produced by the available techniques, their instability on different data sets and the inability to relate statistical significance with biological relevance. Thus, there is an urgent need to address the statistical framework of microarray analysis and identify its drawbacks and limitations, which will enable us to thoroughly compare methodologies under the same experimental set-up and associate results with confidence intervals meaningful to clinicians. In this study we consider gene-selection algorithms with the aim to reveal inefficiencies in performance evaluation and address aspects that can reduce uncertainty in algorithmic validation. RESULTS: A computational study is performed related to the performance of several gene selection methodologies on publicly available microarray data. Three basic types of experimental scenarios are evaluated, i.e. the independent test-set and the 10-fold cross-validation (CV) using maximum and average performance measures. Feature selection methods behave differently under different validation strategies. The performance results from CV do not mach well those from the independent test-set, except for the support vector machines (SVM) and the least squares SVM methods. However, these wrapper methods achieve variable (often low) performance, whereas the hybrid methods attain consistently higher accuracies. The use of an independent test-set within CV is important for the evaluation of the predictive power of algorithms. The optimal size of the selected gene-set also appears to be dependent on the evaluation scheme. The consistency of selected genes over variation of the training-set is another aspect important in reducing uncertainty in the evaluation of the derived gene signature. In all cases the presence of outlier samples can seriously affect algorithmic performance. CONCLUSION: Multiple parameters can influence the selection of a gene-signature and its predictive power, thus possible biases in validation methods must always be accounted for. This paper illustrates that independent test-set evaluation reduces the bias of CV, and case-specific measures reveal stability characteristics of the gene-signature over changes of the training set. Moreover, frequency measures on gene selection address the algorithmic consistency in selecting the same gene signature under different training conditions. These issues contribute to the development of an objective evaluation framework and aid the derivation of statistically consistent gene signatures that could eventually be correlated with biological relevance. The benefits of the proposed framework are supported by the evaluation results and methodological comparisons performed for several gene-selection algorithms on three publicly available datasets.

Epilepsy is one of the most common brain disorders and may result in brain dysfunction and cognitive disturbances. Epileptic seizures usually begin in childhood without being accommodated by brain damage and are tolerated by drugs that produce no brain dysfunction. In this study, cognitive function is evaluated in children with mild epileptic seizures controlled with common antiepileptic drugs. Under this prism, we propose a concise technical framework of combining and validating both linear and nonlinear methods to efficiently evaluate (in terms of synchronization) neurophysiological activity during a visual cognitive task consisting of fractal pattern observation. We investigate six measures of quantifying synchronous oscillatory activity based on different underlying assumptions. These measures include the coherence computed with the traditional formula and an alternative evaluation of it that relies on autoregressive models, an information theoretic measure known as minimum description length, a robust phase coupling measure known as phase-locking value, a reliable way of assessing generalized synchronization in state-space and an unbiased alternative called synchronization likelihood. Assessment is performed in three stages; initially, the nonlinear methods are validated on coupled nonlinear oscillators under increasing noise interference; second, surrogate data testing is performed to assess the possible nonlinear channel interdependencies of the acquired EEGs by comparing the synchronization indexes under the null hypothesis of stationary, linear dynamics; and finally, synchronization on the actual data is measured. The results on the actual data suggest that there is a significant difference between normal controls and epileptics, mostly apparent in occipital-parietal lobes during fractal observation tests.

Investigation of the significance of metabolites peak area ratios derived from brain magnetic resonance spectroscopic imaging (MRSI) spectra, in brain tumors classification, has been applied. Results have shown that in most binary classifications using SVM and LSSVM classifiers, the accuracy achieved was greater than 0.90 AUC except the case of Gliomas grade 2 vs Gliomas grade 3 where 0.84 AUC was recorded due to the great heterogeneity of these two types of tumor. The minimum but also biologically significant set of features (markers), where maximum AUCs recorded, was derived. Ratios of N-acetyl-aspartate, choline, creatine and lipids metabolites found to play the most crucial role in brain tumors discrimination. The biological importance of these markers was also verified by literature. Finally the influence of four magnetic resonance image (MRI) intensities on the classification process was also measured. It was found that MRI data do not improve significantly the classification accuracies.

Following the rapid development of gene selection methods, several comparison studies have been reported for ranking methods on various datasets. In order to reduce bias in performance measures, most studies use an evaluation scheme based on cross-validation. In this paper we focus on the methodology of evaluation itself and address methodological problems using three representative algorithms on two public datasets. More specifically, the paper discusses the need of an independent test-set to reduce bias associated with cross-validation, the use of case specific considerations for generalization, as well as other measures that reflect stability and consistency of the result. Such measures reflect the influence of the actual dataset distribution on the performance of gene selection methods.

The multiple description coding framework is well-suited for the transmission of images in the presence of transmission errors. However, it has been mainly explored for idealized channels, where packets are either received correctly or not at all. In more realistic scenarios, bitstreams are affected by bit-level noise and error correcting codes must be employed. In the proposed scheme, the input image is decomposed into wavelet coefficients that are encoded using lattice vector quantization for the generation of multiple descriptions. Low-density parity-check (LDPC) codes are used for the protection of each stream, while joint decoding exploits the correlation between the multiple descriptions. The proposed joint source-channel decoding scheme can provide robust transmission of images by introducing a controlled amount of redundancy that can be used for error recovery. Results are presented for image transmission over binary symmetric and Gaussian channels.

Information fusion is drawing increasing interest in many application contexts, especially in biomedical decision making. In this work, we provide a framework for addressing the statistical performance of the decision fusion layer. The decision templates (DTs) fusion method is examined as a distance based combiner and statistically compared with an SVM discriminant hyper-classifier. Our aim is broader than providing experimental results on the performance of the two fusion schemes. We attempt to highlight the theoretical advantages of support vectors as multiple attractor points in a hyper-classifier¿s feature space. Moreover we show that the use of SVMs in this task is an extensible framework that can be adapted to the problem formulation.

The problem of marker selection in DNA microarray experiment, due to the "curse of dimensionality", has been mostly addressed so far by linear approaches. Taking into account the fact that the domain of interest is a complex one, where non-linear interconnections and dependencies may also exist among the extremely large number of examined genes, we address the use of nonlinear tools to assess the problem. In this study, we propose to apply the kernel ability of Support Vector Machines in combination with Fisher's ratio as an alternative approach to assess the problem.

Motion information is essential in many computer vision and video analysis tasks. Since MPEG is still one of the most prevalent formats for representing, transferring and storing video data, the analysis of its motion field is important for real time video indexing and segmentation, event analysis and surveillance applications. Our work considers the problem of improving the optical flow field in MPEG sequences. We address the issues of robust, incremental, dense optical flow estimation by combining information from two different velocity fields, the available MPEG motion field and the one inferred by a multiresolution robust regularization technique applied on the DC coefficients. Thus, the regularization technique is based only on information that is directly available in the compressed stream, therefore avoiding time and memory consuming decompression. We extend standard techniques by adding a temporal continuity and an MPEG consistency constraint, both as mathematical constraints in the objective function and as hypothesis tests for the presence of motion discontinuities. Our approach is shown to perform well over a range of different motion scenarios and can serve as a basis for efficient video analysis tasks.

Neural network literature for function approximation is by now sufficiently rich. In its complete form, the problem entails both parametric (i.e., weights determination) and structural learning (i.e., structure selection). The majority of works deal with parametric uncertainty assuming knowledge of the appropriate neural structure. In this paper we present an algorithmic approach to determine the structure of High Order Neural Networks (HONNs), to solve function approximation problems. The method is based on a Genetic Algorithm (GA) and is equipped with a stable update law to guarantee parametric learning. Simulation results on an illustrative example highlight the performance and give some insight of the proposed approach.

In this paper, a hybrid neural network/genetic algorithm technique is presented, aiming at designing a feature extractor that leads to highly separable classes in the feature space. The application upon which the system is built, is the identification of the state of human peripheral vascular tissue (i.e., normal, fibrous and calcified). The system is further tested on the classification of spectra measured from the cell nucleii in blood samples in order to distinguish normal cells from those affected by Acute Lymphoblastic Leukemia. As advantages of the proposed technique we may encounter the algorithmic nature of the design procedure, the optimized classification results and the fact that the system performance is less dependent on the classifier type to be used.

In this paper, a novel framework is proposed to inspect the placement quality of surface mount technology devices (SMDs), immediately after they have been placed in wet solder paste on a printed circuit board (PCB). The developed approach involves the indirect measurement of each lead displacement with respect to its ideal position, centralized on its pad region. This displacement is inferred from area measurements on the raw image data of the lead region through a classification process. To increase the accuracy in the computation of the lead displacement, we introduce a combined classification/estimation process, in which the individual lead displacement classifications are viewed as measurements (or observations) of the same physical quantity i.e., the displacement of the entire component as a rigid body. Certain geometric relations connecting lead shifts to component displacement are also derived. Employing these relations we can infer a new refined measurement of the shift of each individual lead, a quantity crucial to the calculation of the quality measures. Experimental results highlight the potential of the developed algorithm.

A novel framework is proposed to inspect the placement quality of surface mount technology devices (SMDs), immediately after they have been placed in wet solder paste on a printed circuit board (PCB). The considered approach comprises two stages, i.e., observation and Bayesian estimation. The first stage involves the indirect measurement of each lead displacement with respect to its ideal position, centralized on its pad region. This displacement is inferred from area measurements on the raw image data of the lead region through a classification process. To increase the accuracy in the computation of the displacement, the second stage develops a combined classification/estimation process, in which the individual lead displacement classifications are viewed as measurements (or observations) of the same physical quantity i.e., the displacement of the entire component as a rigid body. Experimental results highlight the potential of the developed algorithm.

Assessment of the risk for the development of age related macular degeneration requires reliable detection of retinal abnormalities that are considered as precursors of the disease. A typical sign for the latter are the so-called drusen, which appear as abnormal white-yellow deposits on the retina. This paper presents a novel segmentation algorithm for automatic detection of abnormalities in images of the human eye's retina, acquired from a depth-vision camera. Conventional image processing techniques are sensitive to non-uniform illumination and nonhomogeneous background, which obstructs the derivation of reliable results for a large set of different images. Homomorphic filtering and a multilevel variant of histogram equalization are used for non-uniform illumination compensation and enhancement. We develop a novel segmentation technique, the histogram-teased adaptive local thresholding (HALT), to detect drusen in retina images by extracting the useful information without being affected by the presence of other structures. We provide experimental results from the application of our technique to real images, where certain abnormalities (drusen) have slightly different characteristics from the background and are hard to be segmented by other conventional techniques.

The use of neural network structures for feature extraction and classification is addressed here. More precisely, a nonlinear filter based on higher order neural networks (HONN) whose weights are updated by stable learning laws is used to extract the characteristic features of fluorescence spectra corresponding to human tissue samples of different states. The features are then classified with a multi-layer perceptron (MLP). The high rates of success together with the small time needed to analyze the signals, proves our method very attractive for real time applications.

A wavelet domain robust denoising algorithm is presented, which efficiently removes both Gaussian as well as Gaussian mixed with impulse noise. Several wavelet domain operators are developed which help in the denoising process. The superiority of the new algorithm is firmly established by simulation over a variety of images.

This paper considers the concept of robust estimation in regularized image restoration. Robust functionals are employed for the representation of both the noise and the signal statistics. Such functionals allow the efficient suppression of a wide variety of noise processes and permit the reconstruction of sharper edges than their quadratic counterparts. A new class of robust entropic functionals is introduced, which operates only on the high-frequency content of the signal and reflects sharp deviations in the signal distribution. This class of functionals can also incorporate prior structural information regarding the original image, in a way similar to the maximum information principle. The convergence properties of robust iterative algorithms are studied for continuously and noncontinuously differentiable functionals. The definition of the robust approach is completed by introducing a method for the optimal selection of the regularization parameter. This method utilizes the structure of robust estimators that lack analytic specification. The properties of robust algorithms are demonstrated through restoration examples in different noise environments.

A novel formulation of image processing operations in the wavelet domain is presented which directly associates multiresolution with multichannel image processing. The formation of the multiresolution image is expressed as an operator on the image domain that transforms block-circulant structures into partially-block-circulant structures. The proposed implementation relaxes the stationarity and space-invariance assumptions in the image domain and introduces new operator structures for the implementation of single-channel algorithms which take advantage of the correlation structure in the wavelet domain. Based on this structure, the authors discuss the estimation of the power spectrum in the wavelet domain. Image restoration examples using the linear minimum mean square error filter show significant improvement achieved by the proposed approach over the conventional discrete Fourier transform (DFT) implementation.>

The joint optimization of different criteria is addressed as a powerful means of incorporating prior information in linear image restoration algorithms. The resolution-to-noise-tradeoff (RNT) approach introduced enables the incorporation of both spatial and spectral information regarding the peculiarities of the problem. Depending on the nature of the spatial information, this approach can be interpreted as either a regularized or an adaptive scheme. As a regularized scheme, it offers an alternative to conventional approaches, in which the ringing artifacts, especially those due to the noise, are efficiently suppressed by means of the regularizing functional. As an adaptive scheme, the RNT approach offers the flexibility of applying either linear Wiener filtering, or inverse filtering, or no filtering at all, depending on the local signal-activity. The capabilities of the RNT approach as both a regularized and an adaptive scheme are demonstrated through restoration examples.>

A technique for the design of three-dimensional (3- D) filters is introduced in this paper. The design is based on coefficient transformations of two-dimensional (2-D) circularly symmetric filters. These filters can be designed by cascading 2-D rotated filters. The stability of the 3-D filters designed is discussed and a stabilization procedure based on cepstrum analysis, is proposed. Stable implementation schemes are introduced. Examples of spherically symmetric filters designed on the basis of the technique introduced are presented.