Session MC.5

Enhanced Software Based Detection of Implanted Cardiac Pacemaker Stimuli

M Jennings, B Devine*, S Luo, PW Macfarlane

University of Glasgow
Glasgow, UK

Aims:This study aimed to (i)compare the accuracy of enhanced software, incorporating pattern recognition and edge detection logic for identifying pacemaker spikes using 32K sample/sec (sps) data, against existing software based on current AAMI/IEC standards, in the Atria 6100 ECG machine, using the same data; (ii) assess the ability of the new spike detection logic to aid in the detection of biventricular (BiV) pacing; (iii)develop and test a method for the detection of BiV pacing based on QRS morphology.
Methods: Patients were recruited from pacemaker follow-up clinics or immediately after pacemaker implantation. Each had an ECG recorded using an Atria 6100. Two channels of data at 32Ksps were also obtained and analysed offline by the new spike detection logic, which was developed using a training set of 50 ECGs collected 2 years previously in Glasgow. QRS morphology logic for detecting BiV pacing was developed using 16/50 of these ECGs and the printed Atria 6100 ECGs from the present study were used as a test set for this part of the study. Graphs were produced from the 32Ksps data to assess its potential for capturing (BiV) activity. Statistical analysis was done using difference in proportions for paired data and McNemar’s test.
Results: 72 patients were recruited. 63 were used to assess spike detection accuracy (49 males, 14 females, age 66.1 ± 11.4) and 62 were used to assess the accuracy of reporting BiV pacing (48 males, 14 females, age 66.0 ± 11.5). 33/62 had BiV pacemakers. 9 patients were excluded for various reasons. The new logic improved accuracy of spike detection from 46/63 to 62/63, i.e. by 25% (CI 13%-37%, P<0.001). For BiV pacing, the new logic had a sensitivity of 97% (CI 93%-100%) and a specificity of 100% while the QRS morphology logic had a sensitivity of 70% (CI 58%-81%) and a specificity of 93% (CI 87%-99%).
Conclusion: This study provides strong evidence that the new spike detection logic is significantly more accurate than the old AAMI/IEC standards based logic in the Atria 6100. The new logic is also very efficient in detecting BiV pacing. The QRS morphology logic for reporting BiV pacing may be suitable for less sophisticated types of ECG machines and for personal interpretation of ECGs.

(Abstract Control Number: 26)