Real time monitoring method of cortical somatosensory evoked potential of transform joint sparse model

Abstract

The invention relates to a real time monitoring method of cortical somatosensory evoked potential of a transform joint sparse model. The method comprises following steps of constructing the sparse dictionary of the cortical somatosensory evoked potential; filtering successive two times of electroencephalogram measuring signals by the T-JSM (transform joint sparse model) model through a transform matrix H; carrying out joint sparse decomposition to the filtered electroencephalogram signals, thus obtaining a public sparse coefficient and respective sparse coefficients; reconstructing to obtain single time of CSEP (cortical somatosensory evoked potential) signals; and carrying out real time monitoring, wherein the optimum solutions of the transform matrix H and the sparse coefficients are solved at the same time through adoption of a joint optimization algorithm. According to the method, the single time extraction method of extracting weak CSEP signals from complex intra-operative electroencephalogram signals is broken through; similar components, namely the CSEP signals are separated from two times of electroencephalogram signals; the extraction difficulty is effectively reduced; the extraction accuracy is improved; the transform matrix H is increased; and therefore, the interferences of the EEG (electroencephalogram) to the sparse decomposition of the CSEP signals are reduced.

Description

technical field [0001] The invention belongs to the field of medical signal processing and evoked potential processing, and in particular relates to a real-time monitoring method of cortical somatosensory evoked potential of a transform domain joint sparse model. Background technique [0002] With the rapid development of spinal cord surgery technology, many difficult operations such as scoliosis correction, anterior or posterior decompression of spinal stenosis expansion, intramedullary or extramedullary tumor resection are gradually carried out. The cure rate has also been correspondingly improved. However, some intraoperative operations, such as partial resection of the bony part entering the spinal canal, pedicle screws, implantation of bone grafts, traction and electrocoagulation of the spinal cord during tumor resection, may cause mechanical damage to the normal spinal cord in the corresponding area. Injury and ischemic injury lead to different degrees of spinal cord ...

AI Technical Summary

Problems solved by technology

However, it cannot be done infinitely many times, so the cumulative average method can only improve the signal-to-noise ratio but cannot remove the noise.

In addition, there are two main problems in the cumulative average method: first, the premise of the cumulative average method is to assume that the CSEP signal in each measurement signal is the same and the background noise such as EEG is a random process with a mean value of 0, which does not Inconsistent with the facts, a large number of experiments have proved that there are differences between each CSEP and EEG is non-stationary, which does not fully meet the premise of the cumulative average method; Uninterrupted "accumulation" and "average" are realized, so that the CSEP waveform change after spinal cord injury must be covered by the original waveform, and it can only be gradually manifested after a period of time, which is determined by the accumulation and averaging technology

The problem of the JSR algorithm: EEG signal is a random process with a mean value of 0, and it is impossible to obtain a dedicated sparse dictionary through sample training, and there are overlaps with CSEP in both the time domain and the frequency domain, making some CSEPs easy to be wrongly divided.

Images