Figures of merit to compare distributed linear inverse solutions
- 89 Downloads
This paper introduces the concept of the resolution matrix as the basis for an objective theoretical comparison of distributed linear inverse solutions to the neuroelectromagnetic inverse problem. In particular, we describe how figures of merit derived from the resolution matrices can be represented graphically to evaluate merits and shortcomings of the different solutions. The use of the figures of merit is illustrated with two solutions that consider minimal a priori information about the generators: Classical Minimum Norm and Backus Gilbert. We recommend to start any analysis with the individual exploration of the resolution kernel for each grid point or at least for those points where the activity is likely to occur. This analysis might help in selecting the optimal inverse for the sources that are supposed to be active in the process under study.
Key wordsDistributed solutions Inverse problem Distributed source models Figures of merit
Unable to display preview. Download preview PDF.