Search

In this paper, we propose an end-to-end approach to addressing QRS complex detection and measurement of Electrocardiograph (ECG) paper using convolutional neural networks (CNNs). Unlike conventional detection solutions that convert images to digital data, our method can directly detect QRS complex in images using Faster-RCNN, then the R-peak can be located and measured through a CNN. Validated by clinical ECG data in the St.-Petersburg Institute of Cardiological Technics 12-lead Arrhythmia Database and real ECG paper from Peking University People’s Hospital, the proposed method can achieve the recall of 98.32%, the precision of 99.01% in detecting and 0.012 mv of mean absolute error in measuring. Experimental results demonstrate the superior performance of our method over conventional solutions, which would pave the way to detect and measure ECG paper using CNNs.

In this paper, a novel ergodic cellular automaton model of Hes1 mRNA and Hes1 protein network is presented. Detailed analyses reveal that the presented network model can reproduce a typical nonlinear bifurcation phenomenon observed in a conventional delay differential equation model of the Hes1 mRNA and Hes1 protein network. Furthermore, the presented network model is implemented by a field programmable gate array and its operation is validated by experiments. It is shown that the presented network model consumes much fewer circuit elements and much lower power compared to the delay differential equation network model. Hence the results of this paper will provide fundamental knowledge to design an efficient hardware-based gene network simulator.

Many patients with mental illnesses characterized by impaired cognitive control have no relief from gold-standard clinical treatments resulting in a pressing need for new alternatives. This paper develops a neural decoder to detect task engagement in ten human subjects during a conflict-based behavioral task known as the multi-source interference task (MSIT). Task engagement is of particular interest here because closed-loop brain stimulation during those states can augment decision-making. The functional connectivity patterns of the electrodes are extracted. A principal component analysis of these patterns is carried out and the ranked principal components are used as inputs to train subject-specific linear support vector machine classifiers. In this paper, we show that task engagement can be differentiated from background brain activity with a median accuracy of 89.7%. This was accomplished by constructing distributed functional networks from local field potentials recording during the task performance. A further challenge is that goal-directed efforts take place over higher temporal resolution. Task engagement must thus be detected at a similar rate for proactive intervention. We show that our algorithms can detect task engagement from neural recordings in less than 2 seconds; this can be further improved using an application-specific device.

Muscle synergy is an important method for motor intention recognition in rehabilitation exoskeleton control. The use of the non-negative matrix factorization (NMF) to extract muscle synergy patterns often results in long calculation time due to the amount of data, which makes the effectiveness of synergy extraction low. In this paper, synergy matrices of the complete single-cycle signal while stretching and its segmented ones were extracted respectively. By studying the cosine similarity variation of synergy matrices between each continuous segment and the complete single-cycle EMG signals, it is found that there is a "building-stability-weakening" process on muscle synergy establishment. It is proposed to extract synergy mode with partial data from the "stable" segment, rather than using the complete single-cycle one, as similar result to single-cycle data synergy extraction could be obtained. The calculation time of NMF could be optimized by reducing the amount of data and the real-time characteristics of the synergy mode extraction could be improved at the same time. It is of great significance to use synergy matrix of NMF for motion intention recognition and exoskeleton control.Clinical Relevance- This paper studies the establishment process of the synergy mode, and proposes a method for quickly extracting the synergy mode, which can improve the effectiveness of the recognition of motion intention and is of great significance for the real-time control of the rehabilitation exoskeleton.

This paper focuses on a new algorithm for solving optimization problems using the nature of food search behaviour of caterpillars. The paper describes how the periscopic, pheromonic and fractal search properties analogous to the caterpillars, can aid in designing a new optimization algorithm. The performance characteristics of the new method is compared using 26 standard test functions and the area under the curve of the fitness evaluations is used to validate and compare the proposed algorithms against existing related works. The proposed algorithm is found to be efficient when compared with the existing methods. The proposed algorithm is then tested on a real world problem to remove signal noise from eye gaze data, effectively.

Electrocardiogram (ECG) signal is one of the most important methods for diagnosing cardiovascular diseases but is usually affected by noises. Denoising is therefore necessary before further analysis. Deep learning-related methods have been applied to image processing and other domains with great success but are rarely used for denoising ECG signals. This paper proposes an effective and simple model of encoder-decoder structure for denoising ECG signals (APR-CNN). Specifically, Adaptive Parametric ReLU (APReLU) and Dual Attention Module (DAM) are introduced in the model. Rectified Linear Unit (ReLU) is replaced with the APReLU for better negative information retainment. The DAM is an attention-based module consisting of a channel attention module and spatial attention module, through which the inter-spatial and inter-channel relationship of the input data are exploited. We tested our model on the MIT-BIH dataset, and the results show that the APR-CNN can handle ECG signals with a different signal-to-noise ratio (SNR). The comparative experiment proves our model is better than other deep learning and traditional methods.Clinical Relevance- This paper proposed a method capable of denoising ECG signals with strong noise to alleviate difficulties for further medical analysis.

Passive, continuous monitoring of Parkinson’s Disease (PD) symptoms in the wild (i.e., in home environments) could improve disease management, thereby improving a patient’s quality of life. We envision a system that uses machine learning to automatically detect PD symptoms from accelerometer data collected in the wild. Building such systems, however, is challenging because it is difficult to obtain labels of symptom occurrences in the wild. Many researchers therefore train machine learning algorithms on laboratory data with the assumption that findings will translate to the wild. This paper assesses how well laboratory data represents wild data by comparing PD symptom (tremor) detection performance of three models on both lab and wild data. Findings indicate that, for this application, laboratory data is not a good representation of wild data. Results also show that training on wild data, even though labels are less precise, leads to better performance on wild data than training on accurate labels from laboratory data.Clinical relevance— Results in this paper suggest that, when building a system for in-the-wild PD symptom detection, it is better to train machine learning algorithms on data from the wild as opposed to from the lab, even though wild labels are less precise. This paper also presents a newly released dataset for PD tremor detection in lab and wild environments.

International audience; Different approaches have been proposed in the literature to detect the fall of an elderly person. In this paper, we propose a fall detection method based on the classification of parameters extracted from depth images. Three supervised learning methods are compared: decision tree, K-Nearest Neighbors (K-NN) and Random Forests (RF). The methods have been tested on a database of depth images recorded in a nursing home over a period of 43 days. The Random Forests based method yields the best results, achieving 93% sensitivity and 100% specificity when we restrict our study around the bed. Furthermore, this paper also proposes a 37 days follow-up of the person, to try and estimate his or her daily habits. © 2020 IEEE.

In this paper, a method for the detection and subsequently extraction of neural spikes in an intra-cortically recorded neural signal is proposed. This method distinguishes spikes from the background noise based on the natural difference between their time-domain amplitude variation patterns. According to this difference, a spike mask is generated, which takes on large values over the course of spikes, and much smaller values for the background noise. The "high" part of this mask is designed to be wide enough to contain a complete spike. By multiplying the input neural signal with the spike mask, spikes are amplified with a large factor while the background noise is not. The result is a spike-augmented signal with significantly larger signal-to-noise ratio, on which spike detection is performed much more easily and accurately. According to this detection mechanism, spikes of the original neural signal are extracted.Clinical Relevance—This paper presents an automatic spike detection technique, dedicated to brain-implantable neural recording devices. Such devices are developed for clinical applications such as the treatment of epilepsy, neuro-prostheses, and brain-machine interfacing for therapeutic purposes.

In this paper, we apply combination of sparse representations and a total variation for reconstruction of retinal optical coherence tomography (OCT) images. The OCT imaging is based on interferometry, therefore OCT images suffer from the existence of a high level of noise. Utilization of effective interpolation and denoising algorithms are necessary to reconstruct high-resolution OCT images, especially when the subsampling of data is done during acquisition. In this paper, we take total variational and Morphological Component Analysis (MCA) techniques to reduce noise and interpolate missing data. Different over-complete dictionaries are constructed by using curvelet transform, wavelet transform or DCT, which represent the texture and cartoon layers in B-scans. Comparative analysis of image interpolation is done by two combinations of dictionaries, which are (DCT+Curvelet) and (DWT+Curvelet) transforms. Layered structures are more distinguished in reconstructed image with curvelet dictionary and textures are mostly detectable by wavelet or DCT. Evaluations are done both visually and in terms of different performance measures. Our simulation results show that the (DCT+Curvelet) combination preserve the texture of the image well and the (DWT+Curvelet) combination has better performance in structure preservation.

This paper describes an automatic classification algorithm for nocturnal hypoxemia in patients receiving home oxygen therapy (HOT). Nocturnal hypoxemia is a well-known complication in patients with chronic respiratory disease, and the number of patients receiving HOT has increased in recent years. Many studies have reported that 40% of patients receiving HOT have sleep-related oxygen desaturation. To deal with this situation, a nocturnal pulse oximetry is used to measure oxygen saturation (SpO 2 ) and control the flow rate of highly concentrated oxygen. However, in some cases, the flow rate is not controlled properly and the same flow rate is adopted both during the day and night. There are several types of nocturnal hypoxemia, and it is difficult to classify these types only according to a subjective assessment of a medical doctor. Furthermore, it is difficult to continuously monitor the measurement results of pulse oximetry, although a flexible treatment depending on the state of hypoxemia is desired. To overcome these difficulties, an automatic classification method for SpO 2 measured by the nocturnal pulse oximetry is proposed in this paper. The proposed method uses the time domain waveform and the frequency characteristics of SpO 2 . The classification performance of the method is evaluated by using 48 measured SpO 2 values from patients receiving the HOT. The classification results are validated with decisions of ten chest physicians.

This paper focuses on the data preprocessing scheme, as well as on the frequency selection and spatial filtering modules integrated with a Time-Reversal Multiple Signal Classification (TR-MUSIC) algorithm, for microwave breast imaging. This algorithm is part of the data processing chain of the Wavelia Microwave Breast Imaging (MBI) system prototype, which has been recently installed at the University Hospital of Galway, Ireland, for a first-in-human clinical trial. Indicative results from application of the algorithm on an experimental phantom dataset, and on a first patient dataset, are presented in this paper. Good correspondence between the two datasets is demonstrated, confirming the validity of the experimental setup used so far for the on-site acceptance of the Wavelia system, after installation at the hospital for clinical testing.

Microwave mammography has several advantages e.g. portability, non-contact and cell-friendly measurement, and low cost for the equipment. Therefore, this technology has been regarded as a more frequent screening technique, than the X-ray based modality. This paper focuses on the dependence of the imaging accuracy, which is obtained using inverse scattering algorithms, on the estimation of the breast boundary. The Envelope method has been proposed as a method of using microwaves to estimate the location of the breast boundary. The accuracy of the boundary estimation via the Envelope method depends on that of the time-delay estimation, which is mostly processed by a filter based on waveform matching between the observed and reference signals. However, the coupling effect between the antenna and breast surface deforms the observed waveforms relative to the reference waveform. To mitigate this problem, this paper introduces the finite-difference time-domain (FDTD)-based waveform correction assuming the three-dimensional (3-D) distorted Born iterative method (DBIM) analysis. The 3-D numerical simulations based on realistic breast phantoms demonstrate that the proposed method enhances the accuracy for the 3-D reconstruction of the dielectric profile in highly heterogeneous cases.

Microwave ablation (MWA) is one of the most promising treatment tool to achieve a minimal invasion for human body. For safety and effective ablation for cancerous tissue, the accurate and real-time imaging for a temporal evolution of the ablation zone is highly demanded. This paper assumes the microwave based MWA monitoring, and introduces the novel boundary reconstruction algorithm, which has been demonstrated to achieve a real-time, accurate and noise-robust characteristic. This paper also newly introduces the S 11 based complex permittivity estimator, which is necessary for estimating the ablation boundary. The both finite difference time domain (FDTD) based 3-D numerical test and the experimental investigation demonstrate that the proposed method provides accurate and high-speed 3-D imaging for the ablation zone.

This paper explores the use of ensemble classification methods in the context of the diabetes disease. An analysis was carried out that formulates and answers seven research questions: publication trends, channels and venues; medical tasks undertaken; ensemble types proposed; single techniques used to construct the ensemble methods; rules used to draw the output of the ensemble; datasets used to build and evaluate the ensemble methods; and tools used. A total of 107 papers were chosen after a study selection process. Ensemble methods were applied to diabetes in 2003 for the first time. All medical tasks related to the diabetes disease were investigated, and the diagnosis task was the most frequently addressed activity by means of ensemble methods. The homogeneous ensembles were the most common in the literature. Moreover, decision trees and support vector machines were the most used techniques to build homogeneous and heterogeneous ensembles, respectively. The most frequently found combiner was the majority voting rule. Our findings suggest that ensemble classification methods yield better accuracy than single classifiers. This statement, however, requires an aggregation of the evidence reported in the literature by means of a systematic literature review.

Diabetic retinopathy (DR) is one kind of eye disease that is caused by overtime diabetes. Lots of patients around the world suffered from DR which may bring about blindness. Early detection of DR is a rigid quest which can remind the DR patients to seek corresponding treatments in time. This paper presents an automatic image-level DR detection system using multiple well-trained deep learning models. Besides, several deep learning models are integrated using the Adaboost algorithm in order to reduce the bias of each single model. To explain the results of DR detection, this paper provides weighted class activation maps (CAMs) that can illustrate the suspected position of lesions. In the pre-processing stage, eight image transformation ways are also introduced to help augment the diversity of fundus images. Experiments demonstrate that the method proposed by this paper has stronger robustness and acquires more excellent performance than that of individual deep learning model.

In this paper, we present an evaluation of an adaptation of the Antonyan Vardan Transform (AVT) used in combination with an Extreme Learning Machines (ELM) classifier to process surface electromyography (sEMG) data used to classify six finger movements and a rest state. A total of 12 assays formed by three repetitions performed by four volunteers is analyzed. Additionally, a sample-by-sample output label comparison was performed to make a more comprehensive analysis of the system which was tested on a PC and embedded on a Rasp.berry Pi platform. Compared to literature papers, our system was capable to match or outperform similar solutions even using a simpler model, reaching mean accuracy rates above 94%

In this paper we aim to answer the question, “How can modeling and simulation of physiological systems be used to evaluate life-critical implantable medical devices?” Clinical trials for medical devices are becoming increasingly inefficient as they take several years to conduct, at very high cost and suffer from high rates of failure. For example, the Rhythm ID Goes Head-to-head Trial (RIGHT) sought to evaluate the performance of two arrhythmia discriminator algorithms for implantable cardioverter defibrillators, Vitality 2 vs. Medtronic, in terms of time-to-first inappropriate therapy, but concluded with results contrary to the initial hypothesis- after 5 years, 2,000+ patients and at considerble ethical and monetary cost. In this paper, we describe the design and performance of a Computer-aided Clinical Trial (CACT) for Implantable Cardiac Devices where previous trial information, real patient data and closed-loop device models are effectively used to evaluate the trial with high confidence. We formulate the CACT in the context of RIGHT using a Bayesian statistical framework. We define a hierarchical model of the virtual cohort generated from a physiological model which captures the uncertainty in the parameters and allow for the systematic incorporation of information available at the design of the trial. With this formulation, the estimates the inappropriate therapy rate of Vitality 2 compared to Medtronic as 33.22% vs 15.62% $(\mathrm{p}\lt 0.001)$, which is comparable to the original trial. Finally, we relate the outcomes of the computer- aided clinical trial to the primary endpoint of RIGHT.

Respiration monitoring using microwave Doppler radar has attracted significant interest over the last four decades due to its non-invasive and non-contact form of measurement. However, this technology is still not at the level of practical implementations in healthcare due to motion artifacts and interference from multiple subjects within the range of the Doppler radar sensor. Most reported results in literature focus only on single subject measurements because when multiple subjects are present there are interfering respiration signals which are difficult to separate as individual respiration signals. This paper investigates the feasibility of separating respiratory signatures from the multiple subjects. We employed a new approach using Independent Component Analysis (ICA) with the Joint Approximate Diagonalization of Eignematrices (JADE) algorithm to achieve this for closely spaced subjects, and the system is also capable of estimating Direction of Arrival (DOA) for well-spaced subjects. Experimental results demonstrated that the ICA-JADE method can separate respiratory signatures from two subjects one meter apart from each other at a distance from the radar of 2.89 meters. The separated respiratory pattern closely correlates with reference chest belt respiration patterns, and the mean square error is approximately 11.58%. Concisely, this paper clearly demonstrates that by integrating ICA with the JADE algorithm in a Doppler radar physiological monitoring system, multiple subjects can be monitored simultaneously.

Common Spectral Pattern (CSP) algorithm remains predominant for feature extraction from multichannel EEG motor imagery data. However, multiclass classification of from this featureset has been a challenging job. Different approaches have been proposed to be applied on featureset of different EEG subbands to achieve significant classification accuracy. Ensemble learning is very effective in this context to achieve higher accuracy in Brain-Computer Interface (BCI) domain. In this study, we have proposed enhanced classification algorithms to achieve higher classification accuracies. The methods were evaluated against the motor imagery data from Dataset 2a of the publicly available BCI Competition IV (2008). This dataset consists of 22 channels EEG data of 9 subjects for four different movements. A tree based ensemble approach for supervised classification, Extra-Trees algorithm, has been proposed in this paper and also evaluated for its efficacy on this dataset to classify between left hand and right hand movement imaginations. Moreover, this classifier has its inherent capability to select optimum features. Furthermore, in this paper an extension of the binary classification into multiclass domain is also implemented with error correcting output codes (ECOC) approach using the same dataset. Subject-specific frequency bands $\alpha$ (8-12Hz) and $\beta$ (12-30Hz) along with $HG$ (70-100Hz) were considered to extract CSP features. We have achieved an individual peak accuracy of 98ȥ and 84ȥ in binary class and multiclass classification respectively. Furthermore, the results yielded a mean kappa value of 0.58 across all the subjects. This kappa value is higher than of the winner of competition and also from the most of the other approaches applied in this dataset.

Respiratory rate monitoring is of paramount importance in neonatal care. Manual counting of expansions and contractions of the abdomen or diaphragm of the neonate is still the widely accepted measure of respiratory rate in most clinical settings. A practical, affordable, easy-to-use technology to continuously measure respiratory rate in neonates is essential to recognize the signs and symptoms of respiratory disorders. Clinical validation of a system for continuous and long term respiratory rate monitoring of neonates, in a wearable form factor with the capability of remote monitoring is presented in this paper. The respiratory rate monitor was validated in clinical settings on 10 premature babies with various disease conditions and respiratory rates varying from 25 to 90 breaths per minute. Results show a high degree of correlation between the respiratory rate measured by the device and reference measurements. An intelligent algorithm which can remove motion corruption from the accelerometer data and provide reliable results is essential for large-scale adoption of the technology for both clinical as well as home monitoring. The technical details of implementation, results and analysis of the clinical study and observations made during clinical study regarding the feasibility of integrating the device in neonatal care are covered in this paper.

Clinical neurophysiologists often find it difficult to recall rare EEG patterns despite the fact that this information could be diagnostic and help with treatment intervention. Traditional search methods may take time to retrieve the archived EEGs that could provide the meaning or cause of the specific pattern which is not acceptable as time can be critical for sick neonates. If neurophysiologists had the ability to quickly recall similar patterns, the prior occurrence of the pattern may help make an earlier diagnosis. This paper presents a system that may be used to assist a clinical neurophysiologist in the recall of neonatal EEG patterns. The proposed system consists of an alignment technique followed by an approximate nearest neighbour search algorithm called locality sensitive hashing. The system was tested on six different neonatal EEG pattern types with 430 events in total and the results are presented in this paper.

In this paper, we present a novel Digital Medicines program used for reviewing medication adherence. The program is comprised of an ingestible sensor embedded inside medication and a wearable sensor or patch worn on the skin of the patient. The ingestible sensor activates upon contact with gastric fluids and communicates information about the ingested drug to the patch. Adherence patterns and other physiological markers measured by the system are made available to patients, physicians, and caregivers via mobile and web interfaces. The paper focuses on the wearable sensor hardware and measurement features used to provide a more comprehensive view of the patient’s health centered around and contextualized by adherence patterns. This is achieved using efficient, high-performance signal processing algorithms implemented on a low-power platform. Results from bench and clinical testing are presented to demonstrate the performance of adherence, heart rate, step counts, and body angle measurements.

In the infancy of electromyography (EMG) based pattern recognition (PR) limited numbers of electrode channels were anatomically placed over muscles of interest. Modern methods have shown that regularly spaced electrodes around the circumference of a limb are equally effective and have been demonstrated in consumer-ready myoelectric control systems such as Thalmic Labs’ Myo armband. In addition to linear arrays, grid arrays have also been applied in this field of research. Although electrode arrays have mainly been adopted to simplify placement, other benefits will be exploited in this work.Presented in this paper is a novel spatial-temporal feature set that separately analyzes the intensity and structure of the measured electrical signals (MES) and evaluates the similarities between adjacent electrodes, hence the name Adjacent Features (AF). Results in this paper show that AF produced classification accuracies about 4%-6% greater than autoregression (AR) coefficients and Hudgins’ time-domain (TD) features for classifying 47 hand and wrist gestures, while having a computational simplicity similar to the TD features.

International audience; Minimally Invasive Surgery (MIS) has become widespread as an important surgical technique due to its advantages related to pain relief and short recovery time periods. However, this approach implies the acquisition of special surgical skills, which represents a challenge in the objective assessment of surgical gestures. In this way, several studies shown that kinematics and kinetic analysis of hand movement is a valuable assessment tool of basic surgical skills in MIS. In addition, recent researches proved that human motion performed during surgery can be described as a sequence of constant affine velocity movements. In this paper, we present a novel method to classify gestures based on an affine velocity analysis of 3D motion and an implementation of the Dynamic Time Warping algorithm. In particular, affine velocity calculation correlates kinematics and geometrical variables such as curvature, torsion, and euclidean velocity, reducing the dimension of the conventional 3D problem. In this way, using the simplicity of dynamic time warping algorithm allows us to perform an accurate classification, easier to implement and understand. Experimental validation of the algorithm is presented based on the position and orientation data of a laparoscope instrument, determiMinimally Invasive Surgery (MIS) has become widespread as an important surgical technique due to its advantages related to pain relief and short recovery time periods. However, this approach implies the acquisition of special surgical skills, which represents a challenge in the objective assessment of surgical gestures. In this way, several studies shown that kinematics and kinetic analysis of hand movement is a valuable assessment tool of basic surgical skills in MIS. In addition, recent researches proved that human motion performed during surgery can be described as a sequence of constant affine velocity movements. In this paper, we present a novel method to classify gestures based on an affine velocity analysis of 3D motion and an implementation of the Dynamic Time Warping algorithm. In particular, affine velocity calculation correlates kinematics and geometrical variables such as curvature, torsion, and euclidean velocity, reducing the dimension of the conventional 3D problem. In this way, using the simplicity of dynamic time warping algorithm allows us to perform an accurate classification, easier to implement and understand. Experimental validation of the algorithm is presented based on the position and orientation data of a laparoscope instrument, determined by six cameras. Results show the advantages of the proposed method compared to conventional Multidimensional Dynamic Time Warping to classify surgical gestures in MIS.ned by six cameras. Results show the advantages of the proposed method compared to conventional Multidimensional Dynamic Time Warping to classify surgical gestures in MIS.

With the aging of population, the number of Total Hip Replacement Surgeries (THR) increased year by year. In THR, inaccurate position of the implanted prosthesis may lead to the failure of the operation. In order to reduce the failure rate and acquire the real-time pose of Anterior Pelvic Plane (APP), we propose a measurement system in this paper. The measurement system includes two parts: Initial Pose Measurement Instrument (IPMI) and Real-time Pose Measurement Instrument (RPMI). IPMI is used to acquire the initial pose of the APP, and RPMI is used to estimate the real-time pose of the APP. Both are composed of an Inertial Measurement Unit (IMU) and magnetometer sensors. To estimate the attitude of the measurement system, the Extended Kalman Filter (EKF) is adopted in this paper. The real-time pose of the APP could be acquired together with the algorithm designed in the paper. The experiment results show that the Root Mean Square Error (RMSE) is within 1.6 degrees, which meets the requirement of THR operations.

Nowadays the diagnosis of bladder lesions relies upon cystoscopy examination and depends on the interpreter's experience. State of the art of bladder tumor identification are based on 3D reconstruction, using CT images (Virtual Cystoscopy) or images where the structures are exalted with the use of pigmentation, but none uses white light cystoscopy images. An initial attempt to automatically identify tumoral tissue was already developed by the authors and this paper will develop this idea. Traditional cystoscopy images processing has a huge potential to improve early tumor detection and allows a more effective treatment. In this paper is described a multivariate approach to do segmentation of bladder cystoscopy images, that will be used to automatically detect and improve physician diagnose. Each region can be assumed as a normal distribution with specific parameters, leading to the assumption that the distribution of intensities is a Gaussian Mixture Model (GMM). Region of high grade and low grade tumors, usually appears with higher intensity than normal regions. This paper proposes a Maximum a Posteriori (MAP) approach based on pixel intensities read simultaneously in different color channels from RGB, HSV and CIELab color spaces. The Expectation-Maximization (EM) algorithm is used to estimate the best multivariate GMM parameters. Experimental results show that the proposed method does bladder tumor segmentation into two classes in a more efficient way in RGB even in cases where the tumor shape is not well defined. Results also show that the elimination of component L from CIELab color space does not allow definition of the tumor shape.

Gastroenterology imaging is a diagnostic procedure that incorporates various computer vision challenges for the design of assisted diagnostic systems. The most typical challenge is the design of more adequate visual descriptors that can assist the classification algorithms in getting good diagnostic results. Literature shows that most of the texture descriptors for feature extraction from gastric lesions are based on Gabor filters or local binary patterns (LBP). Although good results are obtained, these techniques have their shortcomings. In this paper, we aim to explore the use of fusion of Gabor filters and LBPs for characterizing gastric lesions. The images are first subjected to Gabor filtering using isotropic Gabor filters, followed by extracting LBPs from the filtered images. We validate the performance of the descriptor on a novel gastroenterology dataset: the Post-MAPS dataset. Our results show that the proposed feature set outperforms the other methods that have been considered in this paper.

In this paper, we present a detailed comparison study of skin segmentation methods for psoriasis images. Different techniques are modified and then applied to a set of psoriasis images acquired from the Royal Melbourne Hospital, Melbourne, Australia, with aim of finding the best technique suited for application to psoriasis images. We investigate the effect of different colour transformations on skin detection performance. In this respect, explicit skin thresholding is evaluated with three different decision boundaries (CbCr, HS and rgHSV). Histogram-based Bayesian classifier is applied to extract skin probability maps (SPMs) for different colour channels. This is then followed by using different approaches to find a binary skin map (SM) image from the SPMs. The approaches used include binary decision tree (DT) and Otsu's thresholding. Finally, a set of morphological operations are implemented to refine the resulted SM image. The paper provides detailed analysis and comparison of the performance of the Bayesian classifier in five different colour spaces (YCbCr, HSV, RGB, XYZ and CIELab). The results show that histogram-based Bayesian classifier is more effective than explicit thresholding, when applied to psoriasis images. It is also found that decision boundary CbCr outperforms HS and rgHSV. Another finding is that the SPMs of Cb, Cr, H and B-CIELab colour bands yield the best SMs for psoriasis images. In this study, we used a set of 100 psoriasis images for training and testing the presented methods. True Positive (TP) and True Negative (TN) are used as statistical evaluation measures.

Screening for breast cancer enables early detection by which curative treatment can be possible. While mammography is the current gold standard for screening, it has low sensitivity in younger women and its harmful X-rays in frequent screening can increase the risk of cancer. Incidence rates are rising in younger women, causing a relook at thermography for low cost and non-harmful screening. In this paper, thermography is compared to mammography correlated with sono-mammography in 65 FNAC/biopsy proven cancer subjects in India. Thermography is comparable to mammography correlated with sono-mammography, having 94% and 95% sensitivity, respectively. A novel semi-automated thermographic tumor detection and location algorithm used in this paper also provides 97% sensitivity. This shows the promise of automated thermographic screening for reaching large populations in a cost effective manner in low resource settings in countries like India. Further studies in a large scale need to be done to evaluate the specificity to enable such solutions.

Brain-Computer Interface (BCI) can provide users with an alternative/augmentative interaction path, based on the interpretation of their brain activity. Steady State Visual Evoked Potentials (SSVEP) paradigm has many appealing features, aiming at implementing BCI-enabled communication-control applications. In this paper, we present a complete signal processing chain for a self-paced, SSVEP-based BCI. The proposed approach mostly focuses at reducing the user effort in dealing with BCI, featuring no need of user-specific calibration or training. In this paper, the classification algorithm is introduced and first validated on offline waveforms, aiming at improving classification accuracy and minimizing the false positive rate. Then, implementation of an online, self-paced SSVEP BCI is illustrated. The scheme refers to a four-way choice and exploits discrimination between intentional control states and nocontrol ones. Good performance is achieved, both in terms of true positive rate (94%), as well as low false positive rate (0.26 min(-1)), even in experiments carried out outside lab-controlled conditions.

In the last few years, there has been an increasing interest in the role of deep fascia mobility in musculoskeletal dynamics and chronic pain mechanisms. In a previous paper we presented an innovative semiautomatic approach to evaluate the 3D motion of the fascia using ultrasound (US) imaging, generating a sparse deformation vector field. This paper presents an improvement of our original method, focusing on the filtering of the sparse vector field and its validation. Moreover, in order to evaluate the performance of the algorithm, a method is proposed to generate synthetic deformation vector fields, including: expansion, rotation, horizontal shear, and oblique shear components. Preliminary tests on the final synthetic deformation vector fields showed promising results. Further experiments are required in order to optimize the tuning of the algorithm.

Accurate assessment of severity of viral respiratory illnesses (VRIs) allows early interventions to prevent morbidity and mortality in young children. This paper proposes a novel imaging biomarker framework with chest X-ray image for assessing VRI's severity in infants, developed specifically to meet the distinct challenges for pediatric population. The proposed framework integrates three novel technical contributions: a) lung segmentation using weighted partitioned active shape model, b) obtrusive object removal using graph cut segmentation with asymmetry constraint, and c) severity quantification using information-theoretic heterogeneity measures. This paper presents our pilot experimental results with a dataset of 148 images and the ground-truth severity scores given by a board-certified pediatric pulmonologist, demonstrating the effectiveness and clinical relevance of the presented framework.

Magnetic Resonance (MR) Fingerprinting is an emerging transient-state imaging paradigm, which enables the quantization of multiple MR tissue parameters in a single experiment. Balanced steady-state free precession (bSSFP)-based MR Fingerprinting has excellent signal-to-noise characteristics and also allows for acquiring both tissue parameter maps and field inhomogeneity maps. However, field inhomogeneity often results in complex magnetization evolutions in bSSFP-based MR Fingerprinting, which creates significant challenges in image reconstruction. In this paper, we introduce a new method to address the image reconstruction problem. The proposed method incorporates a low-dimensional nonlinear manifold learned from an ensemble of magnetization evolutions using a deep autoencoder. It provides much better representation power than a low-dimensional linear subspace in capturing complex magnetization evolutions. We formulate the image reconstruction problem with this nonlinear model and solve the resulting optimization problem using an algorithm based on variable splitting and the alternating direction method of multipliers. We evaluate the performance of the proposed method using numerical experiments and demonstrate that it significantly outperforms the state-of-art method using a linear subspace model.

Wireless Capsule Endoscope (WCE) provides a noninvasive way to inspect the entire Gastrointestinal (GI) tract, including large intestine, where intestinal diseases most likely occur. As a critical component of capsule endoscopic examination, physicians need to know the precise position of the endoscopic capsule in order to identify the position of detected intestinal diseases. Knowing how the capsule moves inside the large intestine would greatly complement the existing wireless localization systems by providing the motion information. Since the most recently released WCE can take up to 6 frames per second, it's possible to estimate the movement of the capsule by processing the successive image sequence. In this paper, a computer vision based approach without utilizing any external device is proposed to estimate the motion of WCE inside the large intestine. The proposed approach estimate the displacement and rotation of the capsule by calculating entropy and mutual information between frames using Fibonacci method. The obtained results of this approach show its stability and better performance over other existing approaches of motion measurements. Meanwhile, findings of this paper lay a foundation for motion pattern of WCEs inside the large intestine, which will benefit other medical applications.

In this paper, we have proposed a low complexity architecture for the Under-determined Blind Source Separation (UBSS) algorithm targeting remote healthcare applications. UBSS algorithm, departing from the typical BSS convention-equal number of the sources and sensors present, which is of tremendous interest in the field of Biomedical signal processing especially for remote health care applications. Since such applications are constrained by the on-chip area and power consumption limitation due to the battery backup, a low complexity architecture needs to be formulated. In this paper, firstly we have introduced UBSS architecture, followed by the identification of the most computationally intensive module N-point Discrete Hilbert Transform (DHT) and finally proposed a low complexity DHT architecture design to make the entire UBSS architecture suitable for such resource constrained applications. The proposed DHT architecture implementation and experimental comparison results show that the proposed design saves 50.28%, 48.40% and 46.27% on-chip area and 53.25%, 48.01% and 45.95% power consumption when compared to the state of the art method for N = 32, 64 and 128 respectively. Furthermore the proposed DHT architecture works for N = 2m point, but the state of art architecture works for N = 4m point, where m is an integer.

Microwave tomography has been proven to successfully reconstruct the dielectric profile of a human breast when used in breast imaging applications, thereby providing an alternative to other imaging modalities. However, the method suffers from high computational requirements which restrict its use in practical imaging systems. This paper presents a novel parallelization strategy to accelerate microwave tomography for reconstruction of the dielectric properties of the human breast. A Time Domain algorithm using this parallelization strategy has been validated and benchmarked against an optimized sequential implementation on a conventional high-end desktop Central Processing Unit (CPU), and a comparison of throughput is presented in this paper. The gain in computational throughput is shown to be significantly higher compared with the sequential implementation, ranging from a factor of 26 to 58, on imaging grid sizes of up to 25 cm square at 1 mm resolution.

This article reports a robust technique for extracting time plane features of Electrocardiogram (ECG) from digital images of ECG paper strips. We concluded this article reporting performance evaluation of the system developed for machine aided cardiac disease detection. Mostly paper based ECG recordings are used in developing countries and digital photographs of different leads could easily be taken and sent with a mediocre cellular phone set. Apart from extracting the features, the proposed system detects cardiac axis deviation and diagnose if Left or Right Bundle Branch Blockage (LBBB or RBBB) is present while fed with the digital photographs of different leads of ECG strips. Preprocessing of the low-resolution images involves background grid line noise removal, adaptive image binarization by Sauvola's method and Bresenham's line joining algorithm to link the ECG signature, if broken. Pattern extraction mainly delineate the time plane features like P wave, QRS complex and T wave using water reservoir based pattern recognition techniques and Discrete Wavelet Transform (DWT). Cardiac axis deviation detection is done by checking the overall voltage levels of QRS complexes of lead I, II and III. Having the knowledge of cardiac axis completes the requirements to comment on the cardiac blockage like Left or Right Bundle Branch Blockage (LBBB or RBBB). Thus, the proposed algorithm is primarily developed for machine aided diagnosis of LBBB or RBBB from the digital photographs of ECG paper strips.

International audience; Huge amounts of surgical data are recorded during video-monitored surgery. Content-based video retrieval systems intent to reuse those data for computer-aided surgery. In this paper, we focus on real-time recognition of cataract surgery steps: the goal is to retrieve from a database surgery videos that were recorded during the same surgery step. The proposed system relies on motion features for video character- ization. Motion features are usually impacted by eye motion or zoom level variations, which are not necessarily relevant for surgery step recognition. Those problems certainly limit the performance of the retrieval system. We therefore propose to refine motion feature extraction by applying pre-processing steps based on a novel pupil center and scale tracking method. Those pre-processing steps are evaluated for two different motion features. In this paper, a similarity measure adapted from Piciarelli's video surveillance system is evaluated for the first time in a surgery dataset. This similarity measure provides good results and for both motion features, the proposed pre- processing steps improved the retrieval performance of the system significantly.

This paper proposes a novel joint time of arrival (TOA)/received signal strength indicator (RSSI)-based wireless capsule endoscope (WCE) location tracking method without prior knowledge of biological human tissues. Generally, TOA-based localization can achieve much higher localization accuracy than other radio frequency-based localization techniques, whereas wireless signals transmitted from a WCE pass through various kinds of human body tissues, as a result, the propagation velocity inside a human body should be different from one in free space. Because the variation of propagation velocity is mainly affected by the relative permittivity of human body tissues, instead of pre-measurement for the relative permittivity in advance, we simultaneously estimate not only the WCE location but also the relative permittivity information. For this purpose, this paper first derives the relative permittivity estimation model with measured RSSI information. Then, we pay attention to a particle filter algorithm with the TOA-based localization and the RSSI-based relative permittivity estimation. Our computer simulation results demonstrates that the proposed tracking methods with the particle filter can accomplish an excellent localization accuracy of around 2 mm without prior information of the relative permittivity of the human body tissues.

This paper reports diffusion weighted MRI measurements of cyclohexane in a novel diffusion tensor MRI phantom composed of hollow coaxial electrospun fibers (average diameter 10.2 μm). Recent studies of the phantom demonstrated its potential as a calibration standard at low b values (less than 1000 s/mm;sup2;/sup) for mean diffusivity and fractional anisotropy. In this paper, we extend the characterization of cyclohexane diffusion in this heterogeneous, anisotropic material to high b values (up to 5000 s/mm;sup2;/sup), where the apparent diffusive motion of the cyclohexane exhibits anomalous behavior (i.e., the molecular mean squared displacement increases with time raised to the fractional power 2α/β). Diffusion tensor MRI was performed at 9.4 T using an Agilent imaging scanner and the data fit to a fractional order Mittag-Leffler (generalized exponential) decay model. Diffusion along the fibers was found to be Gaussian (2α/β=l), while diffusion across the fibers was sub-diffusive (2α/β;l). Fiber tract reconstruction of the data was consistent with scanning electron micrograph images of the material. These studies suggest that this phantom material may be used to calibrate MR systems in both the normal (Gaussian) and anomalous diffusion regimes.

This paper presents a new method for identification of system models that are linear in parametric structure, but arbitrarily nonlinear in signal operations. The strategy blends traditional system identification methods with three modeling strategies that are not commonly employed in signal processing: linear-time-invariant-in-parameters models, set-based parameter identification, and evolutionary selection of the model structure. This paper reports recent advances in the theoretical foundation of the methods, then focuses on the operation and performance of the approach, particularly the evolutionary model determination. The method is applied to the modeling of microbial growth by Monod Kinetics.

The European Society of Cardiology guidelines for heart failure management are based on strong evidence that adherence to optimal medication is beneficial for heart failure patients. Telemonitoring with integrated clinical decision support enables physicians to adapt medication dose based on up to date vital parameters and reduces the number of hospital visits needed solely for up-titration of heart failure medication. Although keeping track of weight and blood pressure changes is recommended during unstable phases, e.g. post-discharge and during up-titration of medication, guidelines are rather vague regarding telehealth aspects. In this paper, we focus on the evaluation of a clinical decision support system for adaption of heart failure medication and for detecting early deteriorations through monitoring of blood pressure, heart rate and weight changes. This clinical decision support system is currently used in INTENSE-HF, a large scale telemonitoring trial with heart failure patients. The aim of this paper was to apply the decision support algorithm to an existing telemonitoring dataset, to assess the ability of the decision support concept to adhere to the guidelines and to discuss its limitations and potential improvements.

This paper presents a new design of worm robot whose body is constructed using a novel crimped elastic mesh braid inspired by the earthworm. The proposed worm robot is intended for inspection within the human body via natural orifices. The design and fabrication procedure of the worm robot are given in the paper. The imitation of peristalsis, used by natural worms, is used to control the worm robot for the purpose of producing motion while causing minimal trauma to biological tissue. The forward locomotive function of the worm robot has been tested on both a flat surface and in a rubber tube. It is shown that the worm robot is capable of propagating forwards for both test conditions in a form similar to the earthworm. The test results indicate the proposed worm robot design has promising application for natural tube inspection, like the colon and the esophagus.

In this paper a methodology to identify sleep apnea events is presented. It uses four easily computable features, three generally known ones and a newly proposed feature. Of the three well known parameters, two are computed from the RR interval time series and the other one from the approximate respiratory signal derived from the ECG using principal component analysis (PCA). The fourth feature is proposed in this paper and it is computed from the principal components of the QRS complexes. Together with a least squares support vector machines (LS-SVM) classifier using an RBF kernel, these four features achieve an accuracy on test data larger than 85% for a subject independent classification, and of more than 90% for a patient specific approach. These values are comparable with other results in the literature, but have the advantage that their computation is straightforward and much simpler. This can be important when implemented in a home monitoring system, which typically has limited computational resources.

This paper proposes a work step indication method for supporting daily work with a grid-pattern projection. To support an independent life of demented senior people, it is desirable that an instruction is easy to understand visually and not complicated. The proposed method in this paper uses a range image sensor and a camera in addition to a projector. A 3D geometry of a target scene is measured by the range image sensor, and the grid-pattern is projected onto the scene directly. Direct projection of the work step is easier to be associated with the target objects around the assisted person, and the grid-pattern is a solution to indicate the spatial instruction. A prototype has been implemented and has demonstrated that the proposed grid-pattern projection is easy to show the work step.

In this paper, a new 3D ultrasound (US) denoising technique that adopts the sparse representation has been proposed for an effective noise reduction in 3D US volumes. The purpose of the proposed method is to reduce image noise while preserving 3D objects edges, hence improving the human interpretation for clinical diagnosis and the 3D segmentation accuracy for further automatic malignancy detection. For denoising 3D US volumes, sparse representation was employed, which has showed an excellent performance in reducing Gaussian noise. It has been well known that US images contain severe multiplicative speckle noise, which has different characteristics compared to the additive Gaussian noise. In this paper, we propose a denoising framework for effectively reducing both Gaussian noise and speckle noise on 3D US volumes. The proposed method removes Gaussian noise using sparse representation. Then, a logarithmic transform is performed to transform the speckle noise into Gaussian noise for applying the sparse representation. To demonstrate the effectiveness of the proposed denoising method, comparative and quantitative experiments had been conducted on a synthesized 3D US phantom data. Experimental results showed that the proposed denoising could improve image quality in terms of denoising measurements.

This paper describes a novel method for the identification of Hammerstein systems with time-varying (TV) static nonlinearities and time invariant (TI) linear elements. This paper develops a linear parameter varying (LPV) state-space representation for such systems and presents a subspace identification technique that gives individual estimates of the Hammerstein components. The identification method is validated using simulated data of a TV model of ankle joint reflex stiffness where the threshold and gain of the model change as nonlinear functions of an exogenous signal. Pilot experiment of TV reflex EMG response identification in normal ankle joint during an imposed walking task demonstrate systematic changes in the reflex nonlinearity with the trajectory of joint position.

This paper discusses concisely the main issues and challenges posed by the validation of retinal image analysis algorithms. It is designed to set the discussion for the IEEE EBMC 2013 invited session “From laboratory to clinic: the validation of retinal image processing tools ”. The session carries forward an international initiative started at EMBC 2011, Boston, which resulted in the first large-consensus paper (14 international sites) on the validation of retinal image processing software, appearing in IOVS. This paper is meant as a focus for the session discussion, but the ubiquity and importance of validation makes its contents, arguably, of interest for the wider medical image processing community.

Scent plays an important role in influencing the brain and has been commonly used in psychological research. Much of such research has been conducted without the use of electroencephalography (EEG) to measure the response of the human brain to scent stimulus. Recent studies have involved the use of EEG to perform comparative studies on how different scents can affect brain activity. However, little has been done to analyze the trend of brain activity when a subject is repeatedly exposed to the same scent. This paper discusses the use of 4 features - Entropy Difference, Entropy Ratio, Entropy Time and Root Mean Square (RMS) to perform trend analysis of EEG signals in a repeated scent-exposure setting. The results show that different types of scents cause the brain to be stimulated at different degrees for each repeated exposure, giving rise to different trend patterns. It is also observed that the 4 features give similar trends for the same scent. This similarity allows us to combine the 4 features by forming a feature vector and plotting them in 3 dimensional (3D) space, using 3 repeated scent exposures as the axes. The region of space where the feature vector lies is represented by an ellipsoid, which can be used to characterize a particular scent. Unlike previous work, which did not characterize scent from EEG recordings, this paper investigates the different trends of scent after its repeated exposure to the human subject and by using the 3D representation to characterize the scent.