Radon transform for adaptive directional time-frequency distributions: Application to seizure detection in EEG signals

Adaptive directional time-frequency distribution (ADTFD) is an efficient TFD, which has outperformed most of the adaptive and fixed TFDs. The ADTFD locally optimizes the direction of the smoothing kernel on the basis of directional Gaussian or double derivative directional Gaussian filter (DGF). However, high computation cost of ADTFD has made this method inconvenient for processing real-life signals, i.e, biomedical signals. This paper addresses this problem and introduces a low-cost ADTFD with much lower computation cost and approximately similar efficiency of ADTFD. In the proposed method, instead of iterative filtering in different directions, which is computationally expensive, the optimized directions are estimated using the Radon transform of the modulus of the signal's ambiguity function. The results show that the proposed method is much faster than the ADTFD.