Detection of pure epileptic transitory among gamma oscillatory activities in MEG signals by SVD, embedded on a partial dynamic reconfiguration

TL;DR

This paper presents a novel method combining SVD and dynamic partial reconfiguration to accurately isolate epileptic transitory activities in MEG signals, significantly improving processing speed and source localization.

Contribution

It introduces an integrated approach of SVD with dynamic reconfiguration to enhance the detection and separation of epileptic transitory signals in MEG data.

Findings

SVD effectively separates transitory activities from gamma oscillations.

The integrated method achieves 27 times faster SVD processing.

Pure epileptic transitory sources are accurately localized.

Abstract

In this work we used the SVD to reconstruct pure epileptic transitory activities for the definition of the accurate sources responsible of the excessive discharge implied by the transitory hallmark activities in pharmaco resistant epilepsy. We applied firstly an automatic detection of the local peaks of the transitory activities through a thresholding the energy distrubition, then we performed the SVD on the detectable transitory (300 ms window) simulated events to recover only the non contaminated transitory activities by gamma oscillations, we calculate the precision of the SVD to evaluate the robustness of this technique in separation between transitory and gamma oscillations. In second phase we applied the SVD on small window of transitory activities from real MEG signal, we obtained dipolar topographic map for the pure transitory activities, however the time excitation for the recover of pure epileptic transitory activities among the gamma oscillations were very heavy and hence we propose to integrate the SVD on a dynamic partial reconfiguration. This integration were very useful, in fact we did obtained about 27 Times faster in the SVD execution.