Prediction of spontaneous termination of atrial fibrillation using time frequency analysis of the atrial fibrillatory wave

In this work, the aim is fo discriminute rerminating from non terminating arrial fibrillation (AF) episodes. A database including a leartiing set with 20 recordings and a test set with 30 recordings was provided. We hypothesized that the most relevant information about the mechanisms that trigger the AF is contained in the atrial fibrillatory wave (FW), which was estimated using a QRST cancellation technique. Spectral and timefrequency analysis of the estimated FW was then pe~ormedP. atients corresponding to the non-terminating group showed a higher frequency than patients from the terminating group, being of 6.73M±0.63Hz and 5.l3±0.34Hz, respectively- A simple threshold on the main frequency classified correctly 27 out of 30 patients in the test set, In addition. two different patterns were observed for $he terminating group. In some patients, it was observed a unique and stable frequency, whereas in other cases it was not possible to track a main frequency, showing an irregular and chaotic pattern.