Detection of Silent Myocardial Ischemia During Ambulatory Monitoring by Time‐Frequency Power Spectral Analysis

Background: The nonstationary conditions of unrestricted ambulatory ECG monitoring preclude any meaningful detection of short‐term transient events by conventional time series methods. Time‐frequency analysis offers a unique approach that tracks the frequency content of the signal as it is changing in time. Methods: We compared the ability of three methods: time‐frequency, RR interval time series, and global HR power spectra to detect the occurrence of silent ischemia (ST depression > 1.5 mm × 80 ms for at least 2 min) among 17 Holter records. The records were obtained from five patients (age 52 ± 2 years) with angiographic coronary artery stenosis > 70%. Individual segments were analyzed in a blinded fashion by three of the authors (MK, EF, AM). Results: Neither changes in the RR interval series nor sequential 2.2‐minute power spectra was significantly predictive of ischemia. Conversely, an abrupt loss of variance in the time‐frequency plot was associated with ST depression 82.3% of the time (Chi square = 3.41; P < 0.055). There was a weaker correspondance between the duration of ST depression and loss of autospectral power. Conclusions: Although a significant correlation (r 2 = 0.863) was seen between the maximum low frequency power (sympathetic modulation) and sudden decreases in RR interval, there were at least three instances where no detectable change in HR occurred during episodes of ischemia. These preliminary observations offer hope that time‐frequency methods may prove useful in the detection of transient events in the ambulatory ECC record.