S4: E-Health 1
- Real-time monitoring of heart rate by processing of Microsoft Kinect™ 2.0 generated streams
Ilaria Bosi, Chiara Cogerino, and Marco Bazzani (Istituto Superiore Mario Boella, Italy)
This paper presents a novel solution for non-invasive real-time heart rate monitoring by the Microsoft device Kinect™ version 2.0. In this paper, it is firstly evaluated the state of the art concerning the solutions for non-invasive detection of vital signs and secondly it is presented the implementation and the tests of a novel methodology for evaluating the heart rhythm by a non-invasive monitoring system based on Kinect™ 2.0 sensor, acting at medium/long distance. The standard method to monitor physiological information exploits photoplethysmographic images, in fact, changes in blood volume can be determined from the spectra of light reflected from (or transmitted through) body tissues. Using a mathematical processing developed and written in Python code, the study shows how it is possible to real-time estimate the heart rate of people in front of the sensor at distance 2.5 m. In order to prove the correctness of the method proposed, two different tests are implemented and 30 different subjects are involved in the test phase. The first test aims to detect the real-time heart rate while the subject is in standing position for 1 minute. The second test gathers heart rate data while subject is performing three different series of rehabilitation exercises. During the tests, each subject wears a pulse oximeter for comparing the values acquired through the Kinect™.
- Cardiac Arrhythmia Detection Using DCT Based Compressive Sensing and Random Forest Algorithm
Tea Marasović, and Vladan Papic (University of Split, Croatia)
The discrimination of ECG signals is of crucial importance in clinical diagnoses of cardiac diseases. Manual analysis of ECG signals is very complex and time consuming task due to their composite nature. This paper proposes a novel scheme for reliable automatic classification of ECG signals into normal and three different abnormal (arrhythmia affected) categories. The feature extraction is based on an amalgamation of discrete cosine transform and random projection for dimensionality reduction. Furthermore, the classification is performed using random forest algorithm. The performance of the proposed scheme is evaluated on the restricted subset of ECG recordings from MIT-BIH arrhythmia database. In the experiments, a near perfect recognition accuracies of 99.33% and 99%, depending on the definition of projection matrix, are achieved with only 50 random projected coefficients; i.e. after considerable dimensionality reduction of the input ECG signal.
- Number of EEG Signal Components Estimated Using the Short-Term Renyi Entropy
Jonatan Lerga and Nicoletta Saulig (University of Rijeka, Croatia); Vladimir Mozetič (Polyclinic Medico, Croatia); Rebeka Lerga (University of Rijeka, Croatia)
Multichannel electroencephalogram (EEG) signals are known to be highly non-stationary and often multi-component. A new method for its complexity, in terms of number of signal components extracted from its time-frequency distributions, has been proposed in this paper. Exploiting its spectral energy variation with time, the joint time-frequency distribution approach was upgraded by the modification of Renyi entropy, called short-term Renyi entropy, and applied to multichannel EEG signal analysis resulting in novel algorithm for its complexity detection. Number of EEG signals components obtained for various EEG signals was shown to provide useful information concerning brain activity at each electrode location, which may further be used to detect the brain activity abnormalities for patients with limb movement difficulties.