Search

CN-121987214-A - Brain signal analysis method based on cross-frequency information fusion

CN121987214ACN 121987214 ACN121987214 ACN 121987214ACN-121987214-A

Abstract

The invention relates to a brain signal analysis method based on cross-frequency information fusion, which is technically characterized by establishing a measurement paradigm based on a brain magnetic diagram and collecting brain magnetic signals, preprocessing the brain magnetic signals, calculating weighted phase lag indexes among different channel signals in preprocessed data, constructing intra-layer connection of a cross-frequency multi-layer network, calculating phase-amplitude coupling among the channel signals in different frequency bands by adopting a modulation index measurement method, constructing inter-layer connection of the cross-frequency multi-layer network, constructing the cross-frequency multi-layer brain function network according to the intra-layer connection and the inter-layer connection, and extracting multi-dimensional characteristics of the cross-frequency multi-layer brain function network. According to the invention, the hierarchical organization structure of the brain function network can be more comprehensively depicted by integrating the connection characteristics of the frequency and the different frequencies at the same time in the cross-frequency domain, the important topological characteristics of the multi-layer network can be stably extracted, the utilization efficiency of the cross-frequency coupling information in the brain magnetic signals can be effectively improved, and the brain signal detection analysis application can be widely realized.

Inventors

  • LU JIEWEI
  • YU NINGBO
  • WU MENGYAN
  • HAN JIANDA
  • WANG JIAYI

Assignees

  • 南开大学
  • 南开大学深圳研究院

Dates

Publication Date
20260508
Application Date
20260121

Claims (8)

  1. 1. A brain signal analysis method based on cross-frequency information fusion is characterized by comprising the following steps: step 1, establishing a measurement paradigm based on a resting state magnetoencephalography and collecting magnetoencephalography signals; step 2, preprocessing the magnetoencephalography signals acquired in the step 1; step 3, calculating weighted phase lag indexes among different channel signals in the preprocessed data, and constructing intra-layer connection of the cross-frequency multi-layer network; step 4, calculating phase-amplitude coupling between channel signals in different frequency bands by adopting a modulation index measurement method, and constructing interlayer connection of a cross-frequency multi-layer network; And 5, constructing a cross-frequency multi-layer brain function network according to the intra-layer connection obtained in the step 3 and the inter-layer connection obtained in the step 4, and extracting multi-dimensional characteristics of the cross-frequency multi-layer brain function network.
  2. 2. The brain signal analysis method based on cross-frequency information fusion according to claim 1, wherein the specific implementation method of the step 1 is that different types of epileptics are selected, and resting brain magnetic signals of the epileptics are acquired at inter-seizure intervals.
  3. 3. The brain signal analysis method based on cross-frequency information fusion according to claim 1, wherein the step 2 is characterized in that the collected brain magnetic signals are preprocessed by the following method: ⑴ The sampling rate of the signal is reduced to 1000Hz; ⑵ Denoising by using a 50Hz notch filter; ⑶ Extracting the interested frequency band by using a band-pass filter of 1-150 Hz; ⑷ Removing the electro-oculogram and electrocardio artifacts by using an independent component analysis technology; ⑸ Data were extracted 300ms before and 350ms after the spike peak.
  4. 4. The brain signal analysis method based on cross-frequency information fusion according to claim 1, wherein the specific implementation method of the step 3 is as follows: Is provided with And Represent the first And Magnetoencephalography signals of channels, two signals of channels are in The phase difference at the moment is: Wherein the method comprises the steps of And Respectively represent brain magnetic signals And At the position of Instantaneous phase obtained by Hilbert transform at moment; the following calculation is performed And Weighted phase lag index between: Wherein, the Representing the mean value on the time scale, The absolute value is represented by a value of, Representing a sign function; The following intra-layer connectivity matrix WM is constructed from the weighted phase lag index: Wherein, the Is the number of channels.
  5. 5. The brain signal analysis method based on cross-frequency information fusion according to claim 1, wherein the specific implementation method of the step 4 is as follows: Is provided with And Indicating the low frequency band Magnetoencephalography signal of channel and the first under high frequency band And (3) performing Hilbert transformation on the magnetoencephalic signals of the channel and the magnetoencephalic signals of the channel respectively: Wherein the method comprises the steps of Representing a Hilbert transform; The following type is used for calculating the low frequency band Instantaneous phase of channel signal and high-band Instantaneous amplitude of channel signal: Will be Dividing into N equal-length phase boxes, and respectively calculating the phase boxes And normalizing: Wherein the method comprises the steps of Is the average amplitude distributed over the mth phase bin; Calculating a modulation index based on KL divergence: where U represents a uniform distribution of data, X represents a data distribution, Is shannon entropy; An interlayer connectivity matrix CM between the following low frequency band L and the high frequency band H is constructed according to the modulation index: Wherein, the Is the number of channels that are to be formed, A signal representing the i-th channel of the low frequency band L, Representing the signal of the jth channel of the high frequency band H.
  6. 6. The brain signal analysis method based on cross-frequency information fusion according to claim 5, wherein the number of the low-frequency band phase bins is set to 18.
  7. 7. The brain signal analysis method based on cross-frequency information fusion according to claim 1, wherein the cross-frequency multi-layer brain function network G constructed in the step 5 is: Wherein, the 、 、 Is provided with a plurality of frequency bands, Is an in-layer connectivity matrix that, Is an inter-layer connectivity matrix.
  8. 8. The brain signal analysis method based on cross-frequency information fusion according to claim 1, wherein said method for multi-dimensional feature extraction in step 5 comprises the steps of: (1) Algebraic connectivity The method is used for representing the information communication degree of the whole cross-frequency multi-layer brain function network: Wherein, the Representing the superlaplace matrix of the multi-layer network, Is an adjacency matrix of a multi-layer network, Is a corresponding degree matrix; (2) Multi-layer clustering coefficients For measuring the strength of the interconnection between a node neighbor: Wherein, the Represent the first Connectivity matrix of layers The number of rows of the device is, The elements of the columns, M is the number of layers, and N is the number of channels; (3) Multiple layer participation coefficients To quantify the coherence of the participation of a node in each network layer: Wherein, the Represent the first Layer node Is defined by the degree of nodes of the (c), Representing nodes And the coincidence node ratio among the multi-layer networks.

Description

Brain signal analysis method based on cross-frequency information fusion Technical Field The invention belongs to the technical field of artificial intelligence, relates to brain signal detection and analysis, and particularly relates to a brain signal analysis method based on cross-frequency information fusion. Background It is well known that the brain has complex cross-frequency coupling processes between neural oscillations in different frequency bands, and is an important mechanism for supporting sensing, motion control, cognitive processing and conscious states. In recent years, a great deal of researches find that the nervous system diseases such as epilepsy, consciousness disturbance, alzheimer disease and the like are accompanied by abnormal multi-band brain function connection and cross-frequency cooperative unbalance, so that the method has important clinical research value for accurately quantifying the cross-frequency information. The brain magnetic graph (MEG) is used as a non-invasive brain function imaging technology with millisecond time resolution, can capture the dynamic changes of different frequency band nerve activities of the brain with high precision, and provides a superior data base for researching cross-frequency coupling. Although MEG signals can provide rich time-frequency characteristics, most of the existing brain function analysis methods are based on static function connection of a single frequency band, and it is difficult to reflect information interaction relations among different frequency bands. For example, the conventional frequency band independent network model can only characterize the same-frequency connection mode, but cannot evaluate the modulation process of the low-frequency rhythm on the high-frequency activity, and cannot systematically describe the multi-frequency band cooperative mechanism. Partial researches introduce local cross-frequency coupling indexes, but the cross-frequency coupling indexes cannot be integrated in a network frame uniformly, so that the influence of cross-frequency coupling on a brain function network structure cannot be described from the whole layer. In addition, the single-layer network structure cannot express interlayer connection, so that modeling capability of the cross-frequency relationship at a network level is severely limited. Therefore, a novel brain network analysis method capable of fusing cross-frequency information and structurally expressing a multi-band connection mode is needed in the brain signal detection analysis technology so as to break through the limitation of the traditional MEG function connection method in cross-frequency modeling, reveal the change characteristics of a multi-band cooperative mechanism from the overall network level and improve the identification capability of related functional abnormalities of the nervous system diseases. Based on the above requirements, it is necessary to establish an analysis method for constructing a cross-frequency multi-layer brain function network by using MEG signals, so as to realize the integrated quantification function of multi-scale neural activity. Disclosure of Invention The invention aims to overcome the defects of the prior art and provides a brain signal analysis method based on cross-frequency information fusion, which expands the functional connection analysis of a traditional single frequency band into a multi-layer and multi-frequency collaborative network structure by introducing cross-frequency coupling and interlayer interaction mapping among various frequency bands and solves the problem that the prior method cannot simultaneously describe the interaction among different frequency bands. The invention solves the technical problems in the prior art by adopting the following technical scheme: A brain signal analysis method based on cross-frequency information fusion comprises the following steps: step 1, establishing a measurement paradigm based on a resting state magnetoencephalography and collecting magnetoencephalography signals; step 2, preprocessing the magnetoencephalography signals acquired in the step 1; step 3, calculating weighted phase lag indexes among different channel signals in the preprocessed data, and constructing intra-layer connection of the cross-frequency multi-layer network; step 4, calculating phase-amplitude coupling between channel signals in different frequency bands by adopting a modulation index measurement method, and constructing interlayer connection of a cross-frequency multi-layer network; And 5, constructing a cross-frequency multi-layer brain function network according to the intra-layer connection obtained in the step 3 and the inter-layer connection obtained in the step 4, and extracting multi-dimensional characteristics of the cross-frequency multi-layer brain function network. Further, the specific implementation method of the step 1 comprises the steps of selecting different types of epileptics and collec