Pulse wave decomposition and analysis in diabetic patients using Gaussian peak fitting / by Sebastian Schwarzenberger
VerfasserSchwarzenberger, Sebastian
Begutachter / BegutachterinKaniusas, Eugenijus ; Kampusch, Stefan
Umfang56 S. : graph. Darst.
HochschulschriftWien, Techn. Univ., Mag.-Arb., 2015
Zsfassung in dt. Sprache
Schlagwörter (EN)Pulse wave / diabetis / optical plethysmography
URNurn:nbn:at:at-ubtuw:1-91111 Persistent Identifier (URN)
 Das Werk ist frei verfügbar
Pulse wave decomposition and analysis in diabetic patients using Gaussian peak fitting [1.91 mb]
Zusammenfassung (Englisch)

Diabetes induced complications like peripheral vascular disease (PVD) and neuropathy promote the formation of wounds and lead to impaired wound healing in diabetic patients. Early detection and effective, low cost treatment options are key factors to improve the patient's quality of life as well as to reduce health care costs. New therapeutic approaches like percutaneous auricular vagus nerve stimulation (aVNS) promise to reactivate the autonomous nervous system (ANS), promote wound healing or prevent the formation of new wounds. To monitor the effects of aVNS on the ANS, the pulse plethysmography (PPG) signal of healthy and diabetic patients was recorded and analyzed in the course of a pilot study. Parameters of interest were the systolic-diastolic volume deflection, the reflection time and the reflection index. For the latter two, the respective incident and reflected pulse waves in each cardiac cycle need to be extracted from the PPG signal. This was achieved by Gaussian peak fitting, using an iterative least-square fitting algorithm. The algorithm was validated by the use of artificial and real measured pulse signals. For most regular PPG signals, fitting errors of < 5 % could be achieved. Using appropriate quality control mechanisms, highly automated and accurate analysis is possible. Besides a decomposition of incident and reflected waves, the original PPG signal can also be reconstituted from the Gauss parameters, and thus may help to save memory. The feasibility of analyzing full datasets from the pilot study was tested. In this analysis, differences between single healthy and diabetic patients could be observed. Thus, the presented method holds potential for diagnostics and a possible monitoring of therapeutic effects. Long-term effects were not assessed and further analysis is necessary.