Analysis of the electrical activity organization within the atria can reveal interesting clinical information related to atrial fibrillation (AF), which is the most commonly diagnosed arrhythmia in clinical practice. However, a very reduced number of indirect AF organization estimators from the surface electrocardiogram (ECG) have been proposed to date. The present work introduces a new method for direct and short-time AF organization estimation from each surface lead of ECG recordings. The temporal arrhythmia organization was estimated through the computation of morphological variations among single fibrillatory (
) waves. They were delineated and extracted from the atrial activity (AA) signal using an adaptive signed correlation index. The algorithm was tested on real AF recordings to discriminate atrial signals with different organization degrees. Results indicated a notably higher global accuracy (90.3%) in comparison with the two non-invasive AF organization estimates widely used today: the dominant atrial frequency (70.5%) and sample entropy (76.1%). Moreover, due to its ability to assess AA regularity wave-to-wave, the proposed method was also able to pursue AF organization time course more precisely than the aforementioned indices.