Yuriy Kurylyak

Photoplethysmography detection by smartphone's videocamera

Smartphones video cameras become more and more powerful while the devices itself become used by many people. That allows utilizing them for many every-day tasks. One of such suitable application of the widely used smartphones is monitoring the state of the health. In this paper we propose an approach to detect the photoplethysmograph signal from the fingertip using a smartphones camera and built-in LED. The proposed solution allows detecting the proper heart rate and is robust to different situations of wrong-usage of the system.

Growing gold nanoparticles on a flexible substrate to enable simple mechanical control of their plasmonic coupling

A simple method is presented to control and trigger the coupling between plasmonic particles using both a growing process of gold nanoparticles (GNPs) and a mechanical strain applied to the elastomeric template where these GNPs are anchored. The large scale samples are prepared by first depositing and then further growing gold nanoparticles on a flexible PDMS tape. Upon stretching the tape the particles move further apart in the direction of the stretching and closer together in the direction perpendicular to it. The synergy between the controlled growth of GNPs and the mechanical strain, leads to a drastic shift of the plasmon band and a color change of the sample. Furthermore, the stretching by only a few percent of the amorphous and initially isotropic sample results in a strong polarization-dependent plasmon shift. At smaller gap sizes between neighboring particles, induced by stretching the PDMS tape, the plasmon shift strongly deviates from the behaviour expected considering the plasmon ruler equation. This shows that multipolar coupling effects significantly contribute to the observed shift. Overall, these results indicate that a macroscopic mechanical strain allows one to control the coupling and therefore the electromagnetic field at the nanoscale.

A Neural Network-based method for continuous blood pressure estimation from a PPG signal

There is a relation, not always linear, between the blood pressure and the pulse duration, obtained from photoplethysmography (PPG) signal. In order to estimate the blood pressure from the PPG signal, in this paper the Artificial Neural Networks (ANNs) are used. Training data were extracted from the Multiparameter Intelligent Monitoring in Intensive Care waveform database for better representation of possible pulse and pressure variation. In total there were analyzed more than 15000 heartbeats and 21 parameters were extracted from each of them that define the input vector for the ANN. The comparison between estimated and reference values shows better accuracy than the linear regression method and satisfy the American National Standards of the Association for the Advancement of Medical Instrumentation.

Detection of the eye blinks for human's fatigue monitoring

This paper presents a non-intrusive vision based system for eye blinks detection and fatigue level monitoring. It uses a web camera positioned in front of the face. A cascade of boosted classifiers based on Haar-like features is used for fast detection of the eyes region. The frames differencing in combination with the thresholding are applied to detect the eyes closure and opening. The frame processing algorithm is pointed out in order to distinguish the involuntary blinks from the voluntary ones. Experimental tests are shown that validate the proposed system.

Reliable pulse rate evaluation by smartphone

The paper deals with the robust and reliable evaluation of the Pulse Rate (PR) with a smartphone. The smartphone camera is used to evaluate the volumetric variation of blood by monitoring the change of light absorption in the tissue. Once assessed the correct working operation, the PhotoPlethysmoGram (PPG) signal is detected and the PR is evaluated on the basis of adaptive and statistical analysis. To validate the pointed out method, the PR, evaluated by smartphone, is compared with the Ambulatory Blood Pressure monitor Spacelabs 90207, that is clinically validated medical device. The experimental results confirm the correctness and suitability of the proposed method.

Application of the Artificial Neural Network for blood pressure evaluation with smartphones

The smartphone is proposed to evaluate the Blood Pressure (BP) anywhere and anytime. The tasks performed by smartphone are (i) extraction of the PhotoPlethysmoGram (PPG) signal from a frame sequence acquired by the integrated camera, and (ii) processing it by Artificial Neural Network for the evaluation of the BP. The PPG signal is evaluated by analyzing the volumetric blood variation of the fingertip on the frame sequence. Successively, parameters characterizing the pulses of the PPG signal are sent to the Fit Forward Neural Network for the simultaneously evaluation of the systolic and the diastolic BP. The validation of the results is performed by comparing them with the ones obtained by the Ambulatory Blood Pressure monitor ABP Spacelabs 90207. Preliminary experimental results show useful information to address the future research devoted to reduce the maximum error.

Logistic ordinal regression for the calibration of oscillometric blood pressure monitors

Abstract Oscillometric blood pressure (BP) monitors are omnipresent and used on a daily basis for personalized healthcare. Nevertheless, physicians generally approach these devices cautiously since the mercury Korotkoff sphygmomanometer remains the golden standard. Various reasons explain the hesitating attitude of the medical world towards automated BP monitors: (i) its principle is based on the pressure pulsations arriving at the cuff by the cardiac cycle instead of an audio wave used by physicians triggered by the turbulences in the artery, (ii) the actual computation of the systolic and diastolic BP from the measured oscillometry is manufacturer dependent and not based on general scientific principles, (iii) the quality of the oscillometric monitors is labeled by a trial such that the devices correspond well to the Korotkoff method for the average healthy patient but deviates for patients suffering from hypo- or hypertension. In this paper, we develop a statistical learning technique to calibrate and correct an oscillometric monitor such that the device better corresponds to the Korotkoff method regardless of the health status of the patient. The technique is based on logistic regression which allows correcting and eliminating systematic errors caused by patients suffering from hyper -or hypotension. No user interaction is required since the technique is able to train and validate the calibration procedure in an unsupervised way. In our case study, the systematic error is reduced by nearly 50% corresponding to the performance specifications of the device.

Photoplethysmogram-based Blood pressure evaluation using Kalman filtering and Neural Networks

The paper deals with the accurate evaluation of the Blood pressure (BP) by an Artificial Neural Network (ANN) and the Photoplethysmogram (PPG) signal. The proposed method allows evaluating the blood pressure for each heart beat, without using a cuff or invasive tool. For each heart beat, a fixed number of features, which characterize the PPG pulse, are extracted and given as the input to the ANN. A systolic, diastolic and mean BP are obtained as the output. The improvement of the BP evaluation accuracy is obtained by removing artifacts from the references used to train the ANN. The filtering of the reference inputs is performed with Kalman based filter in order to take into account the variability of the human pulse rate and cardiovascular system. Preliminary experimental results confirm the suitability of the proposal and asses the BP evaluation accuracy within 5±8 mmHg.

Using the heart harmonics in the oscillometry to extract the blood pressure

Non-invasive measurement of the human blood pressure remains, although the commercial availability of such monitors, an active research topic. This is due to the disadvantage that the blood pressure is computed instead of directly measured. To compute the systolic and diastolic blood pressures from the oscillometry, the envelope of the oscillometric waveform is used to retrieve the systolic and diastolic pressures. This algorithm to determine the blood pressure depends on the brand of oscillometric device and as a result, this type of blood pressure measurement is scientifically ill-founded. In this paper, we investigate the possibility of extracting the systolic and diastolic blood pressures from the hearts fundamental frequency and its harmonics. This signals spectrum shows similarities to the spectrum of the actual arterial blood pressure signal which is observable by invasive measurements. The algorithm we present to obtain the systolic and diastolic blood pressures from the heart harmonics operates with the same accuracy as commercial devices. Finally the new technique can be used to calibrate commercial oscillometric devices where we show an improvement of 4 mmHg.

Detection and parameters estimation of locally motion blurred objects

This paper presents a new approach to detect moving objects affected by motion blur. The direction and the length of the blur reflect the original motion of the object during a time of pictures acquiring by the camera (shutter speed). The analysis of image in the spectrum domain using Discrete Cosign or Fourier Transforms allows detecting of the motion blurs direction and speed from the image. However, such techniques do not work when there are a few objects on the image with different blur or complex background. The proposed method of local motion bluer detection, based on image partitioning, allows locating only of the regions affected by the motion blur and, therefore, measuring the motion parameters of multiple objects.