KMITL

Permanent URI for this communityhttps://dspace.kmitl.ac.th/handle/123456789/1

Browse

Settings

Author: search.filters.author.Sinchai, SakkarinSubject: search.filters.subject.Amplitude modulation (AM)

Search Results

Now showing 1 - 1 of 1
  • Loading...
    Thumbnail Image
    Some of the metrics are blocked by your 
    consent settings
    Item type:Publication,
    Performance of frequency translation in separating a photoplethysmographic signal from an additive motion artifact
    (2019-02-01)
    Koseeyaporn, Jeerasuda
    ;
    Sinchai, Sakkarin
    ;
    Wardkein, Paramote
    Obtaining red and infrared (IR) photoplethysmographic (PPG) signals by a general method in a commercial pulse oximeter is overlapped with an additive motion artifact (MA) signal routinely. When the red and IR PPG signals are combined with the MA signal, a percentage of oxygen saturation (SpO2) is unreliable. To prevent the overlapping problem, a technique of frequency translation is introduced to shift the PPG frequency components away from the MA frequency components. The introduced approach remodels an LED-driving system by substituting an alternating current (AC) source for a traditional direct current (DC) source in the commercial pulse oximeter. To assess the performance, the SpO2 values computed from the red and IR PPG signals acquired by the presented solution are evaluated when four natural poses of motion occur. Besides, the well-known methods are used to calculate the SpO2 values from the red and IR PPG signals sensed by the conventional LED-emitting system during motion for the efficient comparison. The well-known methods are discrete saturation transform (DST), fast independent component analysis (fICA) and compression of Fourier coefficients (CFC). The resulting SpO2 values show that the technique of frequency translation provides overall mean error lower than the selected schemes for all postures. The overall mean error of the introduced technique is 1.1% while the approaches of DST, fICA and CFC yield the overall mean errors by 2.9%, 15.4% and 8.4%, respectively.