Presentation is loading. Please wait.

Presentation is loading. Please wait.

Tayseer O’Brien Dr. R. Eric McGregor INTRODUCTION

Similar presentations


Presentation on theme: "Tayseer O’Brien Dr. R. Eric McGregor INTRODUCTION"— Presentation transcript:

1 Carotid Pulse Detection and Signal Enhancement in a Biometric Identification System
Tayseer O’Brien Dr. R. Eric McGregor INTRODUCTION The existing MatLab software program named Thoracic Identification System (TIS), written by Dr. R. Eric McGregor for the project titled Thoracic Biometric –Investigations of the Human Heartbeat as a Biometric: Heart Sounds, Electrocardiogram, and Vibrometry, investigated the use of traits from the human heart as a biometric that cannot be easily disguised. The program demonstrated good performance on Electrocardiogram(ECG) data but not on the Carotid Pulse(CP) data. This project focused on improving the performance on the CP data by modifying the R-Peak Heartbeat Detection and Heartbeat Segment Selection algorithms. These modifications significantly increased the overall performance from 27.27% to 60.61%. MODIFICATION#1  R-Peak Heartbeat Detection Algorithm Algorithm rewrite A window size, that is set by user, traverses through the heartbeats and segments them by detecting the maximum value (R-Peak) The window size determines the least distance between each possible R-Peak found Peaks are missed but there are no false-positives TEST RESULTS Headings: 40 Ariel Type: 32 J: 20 Veranda for text 36 Heading Big for title Before Biometric Identification System After MODIFICATION#2  Heartbeat Segment Selection Algorithm Original Algorithm – chooses k random beats or k consecutive beats Improved Algorithm - picks k beats that are closest to the subject’s nominal (average) heartbeat unique heartbeat segments that are different from the rest of the subjects EXPERIMENTAL DATA 36 total subjects recordings trials from each Trials 1-4  at rest Trials 5-8  after exercise Trials 9-12  5 minutes after exercise Collected the following cardiac data: Carotid Pulse (CP) using Laser Doppler Vibrometer (LDV) Electrocardiogram (ECG) using electrodes placed on subjects’ thoracic region Phonocardiogram (PCG) using a digital stethoscope CONCLUSION This project allowed the matching performance of TIS to significantly increase from 27.27% to 60.61% in identifying subjects based on their CP data when training on rested heartbeats from recording 1 and testing on elevated heartbeats in recording 9. Modifying the Heartbeat Detection and the Heartbeat Segment Selection algorithms made it more possible for the program to see the R-Peaks for segmentation purposes and to determine which segments are mostly unique to that subject. BIBLIOGRAPHY R.E.McGregor; S.Schuckers; J.Skufca, "Thoracic Biometrics - Investigations of the Human Heartbeat as a Biometric: Heart Sounds, Electrocardiogram, and Vibrometry," CIA Report (not published), 2013. M. Chen; J.A. O'Sullivan; N. Singla; E.J. Sirevaag; S.D. Kristjansson; P.H. Lai; A.D. Kaplan; J.W. Rohrbaugh; , "Laser Doppler Vibrometry Measures of Physiological Function: Evaluation of Biometric Capabilities," IEEE Transactions on Information Forensics and Security, v..5, no.3, pp , 2010. M. Chen, “Laser Doppler Vibrometry as Biometric”, M.S. Thesis, Washington University, 2007. Select recordings for training and testing Training Identify R-Peaks Construct spectrogram Heartbeat segment selection Classification (Matching) Testing Identify R-Peaks Construct spectrogram Heartbeat segment selection


Download ppt "Tayseer O’Brien Dr. R. Eric McGregor INTRODUCTION"

Similar presentations


Ads by Google