Real-time recomputation of the linear inverse operator from updated noise covariance. More...


Go to the source code of this file.
Classes | |
| struct | RTPROCESSINGLIB::RtInvOpInput |
| Input bundle for the real-time inverse operator worker containing noise covariance, forward solution, and settings. More... | |
| class | RTPROCESSINGLIB::RtInvOpWorker |
| Background worker thread that recomputes the MNE inverse operator when covariance updates arrive. More... | |
| class | RTPROCESSINGLIB::RtInvOp |
| Controller that manages RtInvOpWorker for online inverse operator updates. More... | |
Namespaces | |
| namespace | FIFFLIB |
| FIFF file I/O, in-memory data structures and high-level readers/writers. | |
| namespace | MNELIB |
| Core MNE data structures (source spaces, source estimates, hemispheres). | |
| namespace | RTPROCESSINGLIB |
Real-time recomputation of the linear inverse operator from updated noise covariance.
SPDX-License-Identifier: BSD-3-Clause Copyright (c) 2026 MNE-CPP Authors
RtInvOp recomputes a regularised MNELIB::MNEInverseOperator whenever a new noise-covariance estimate is published by RtCov, so a live source-localisation pipeline can adapt to changes in the sensor noise floor (subject movement, ambient electromagnetic noise, channel removal) without dropping a sample. The forward solution, source-orientation settings and SNR / depth-prior parameters are held constant; only the covariance-dependent factors are reassembled.
The actual whitening, SVD and assembly run on a worker QThread; the resulting inverse operator is delivered via a Qt signal so downstream applyInverse calls see a consistent operator at all times.
Definition in file rt_inv_op.h.