IMPROVING THE DIAGNOSIS OF ACUTE MYOCARDIAL INFARCTION BY DERIVING EPICARDIAL POTENTIALS FROM THE BODY SURFACE POTENTIAL MAP USING INVERSE ELECTROCARDIOGRAPHY AND AN INDIVIDUALIZED TORSO MODEL

Abstract

Epicardial potentials (EP) derived from the body surface potential map (BSPM) improve acute myocardial infarction (AMI) diagnosis. In this study, we compared EP derived from the 80-lead BSPM using a standard thoracic volume conductor model (TVCM) with those derived using a patient-specific torso model (PSTM) based on body mass index (BMI). Patients presenting to ED between August 2009 and August 2011 with acute ischaemic-type chest pain at rest were enrolled. At first medical contact, 12-lead ECG and BSPM were recorded. BMI for each patient was calculated. Cardiac troponin-T (cTnT) was sampled 12h after symptom onset. Patients were excluded from analysis if they had any electrocardiographic confounders to interpretation of the ST-segment. A cardiologist assessed the 12-lead ECG for STEMI by Minnesota criteria and BSPM. BSPM ST-elevation (STE) was ≥0.2mV in anterior, ≥0.1mV in lateral, inferior, RV or high right anterior and ≥0.05mV in posterior territories. To derive EP, the BSPM data were interpolated to yield values at 352-nodes of a Dalhousie torso. Using an inverse solution based on the boundary element method, EP at 98 cardiac nodes positioned within a standard TVCM were derived. The TVCM was then scaled to produce a PSTM, using a model developed from CT in 48 patients of varying BMI, and EP re-calculated. EP ≥0.3mV defined STE. A cardiologist blinded to both the 12-lead ECG and BSPM interpreted the EP map. AMI was defined as cTnT ≥0.1µg/L. Enrolled were 400 patients (age 62 ± 13 yrs; 57% male): 80 patients had exclusion criteria. Of the remaining 320 patients, BMI was 27.8 ± 5.6kg m-2. Of these, 180 (56%) had AMI. Overall, 132 had Minnesota STEMI on ECG (sensitivity 65%, sensitivity 89%) and 160 had BSPM STE (sensitivity 81%, specificity 90%). EP STE occurred in 165 patients using TVCM (sensitivity 88%, specificity 95%, p<0.001) and in 206 patients using PSTM (sensitivity 98%, specificity 79%, p<0.001). Of those with EP <0.3mV using TVCM and AMI (n=22), all had EP ≥0.3mV when an individualised PSTM was used. Among patients presenting with ischaemic-type chest pain at rest, EP derived from BSPM using a novel PSTM significantly improves sensitivity for AMI diagnosis.