Compressed sensing of large-scale local field potentials using adaptive sparsity analysis and non-convex optimization

摘要

Objective. Energy consumption is a critical issue in resource-constrained wireless neural recordingapplications with limited data bandwidth. Compressed sensing (CS) has emerged as a powerfulframework in addressing this issue owing to its highly efficient data compression procedure. In thispaper, a CS-based approach termed simultaneous analysis non-convex optimization (SANCO) isproposed for large-scale, multi-channel local field potentials (LFPs) recording. Approach. TheSANCO method consists of three parts: (1) the analysis model is adopted to reinforce sparsity ofthe multi-channel LFPs, therefore overcoming the drawbacks of conventional synthesis models.(2) An optimal continuous order difference matrix is constructed as the analysis operator,enhancing the recovery performance while saving both computational resources and data storagespace. (3) A non-convex optimizer that can by efficiently solved with alternating direction methodof multipliers is developed for multi-channel LFPs reconstruction. Main results. Experimentalresults on real datasets reveal that the proposed approach outperforms state-of-the-art CS methodsin terms of both recovery quality and computational efficiency. Significance. Energy efficiency ofthe SANCO make it an ideal candidate for resource-constrained, large scale wireless neuralrecording. Particularly, the proposed method ensures that the key features of LFPs had littledegradation even when data are compressed by 16x, making it very suitable for long term wirelessneural recording applications.