Determination of Electrode Exchange for 12-Lead ECG Based on Correlation Coefficient Scores

The exchange of electrodes during ECG testing can lead to misdiagnosis in up to 24% of cases, some of which are severe medical conditions. To solve this problem, we develop an algorithm based on the exchanges of the elements of the 6x6 correlation coefficient score matrix for precordial leads (V1-V6) and maximum correlation with the V6 electrode for peripheral leads (I, II, III, -I, -II, -III, -aVR, -aVF and - aVL). The algorithm was validated using the k-fold crossvalidation technique (k=8) for 545 records from the PTB database. The algorithm analyzes specific correlation coefficients of simulated exchanges to minimize the computational cost. The results report average Sensitivity (Se) and Specificity (Sp) for the algorithm finding values of Se= 99.52% and Sp= 99.62% for one-, two- and three-hop precordial electrode exchanges, and Se= 92.10%, and Sp= 96.05% for exchanges in peripheral leads. The electrode exchange detection algorithm based on inter-lead correlation analysis provides accurate detection in both precordial and peripheral electrode exchanges in standard 12-lead ECGs, thus validating its implementation in biomedical devices such as the Think Health project.