Cortical layer signal reconstruction method based on physical constraint diffusion model and related equipment

By using a physical constraint diffusion model, ECoG is reconstructed from EEG signals, which solves the problems of missing high-frequency details and insufficient physical constraints in EEG-to-ECoG reconstruction, and achieves accurate recovery of high-frequency signals and physically interpretable mapping.

CN122132681APending Publication Date: 2026-06-02SUN YAT SEN UNIV
0 Cites 1 Cited by

Patent Information

Application Number
CN202610136853.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-01-30
Publication Date
2026-06-02

AI Technical Summary

Technical Problem

Existing technologies struggle to accurately reconstruct invasive electrocorticography (ECoG) using non-invasive electroencephalography (EEG), particularly in terms of recovering high-frequency signal details and adhering to the physical laws of electromagnetic propagation.

Method used

A method based on a physical constraint diffusion model is adopted to extract conditional feature vectors from EEG signals. Then, using a trained diffusion model and a physical sensing variational autoencoder, combined with the lead field matrix, generative denoising and signal reconstruction are performed to ensure that the generated signal conforms to the electromagnetic propagation law.

Benefits of technology

It effectively restores high-frequency neural oscillation details that are attenuated or lost due to the skull low-pass filtering effect, achieves accurate mapping from EEG to ECoG, and improves the spatial resolution and clinical reliability of the signal.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention discloses a cortical signal reconstruction method based on a physical constraint diffusion model and related equipment, and the method comprises the steps: obtaining a non-invasive EEG signal of the head of a target object, and extracting a condition feature vector according to the non-invasive EEG signal; inputting the condition feature vector into a trained diffusion model for denoising to obtain a transient latent variable, and obtaining a transient ECoG estimated value according to the transient latent variable and the trained physical perception variational auto-encoder; calculating a theoretical signal according to the transient ECoG estimated value and a preset lead field matrix, and updating a transient latent variable according to the difference between the theoretical signal and the noninvasive EEG signal until a target latent variable is obtained; the preset lead field matrix represents a transfer relation from a cortex source space to a scalp sensor space; and obtaining an ECoG reconstruction signal according to the target latent variable and the trained physical perception variational auto-encoder. According to the embodiment of the invention, the ECoG can be accurately predicted through the EEG. The method can be widely applied to the technical field of electroencephalogram.
Need to check novelty before this filing date? Find Prior Art