A weighted least-squares algorithm for estimation and visualization of relative latencies in event-related functional MRI.

The properties of the hemodynamic latencies in functional maps have been relatively unexplored. Accurate methods of estimating hemodynamic latencies are needed to take advantage of this feature of fMRI. A fully automated, weighted least-squares (WLS) method for estimating temporal latencies is reported. Using a weighted linear model, the optimal latency and amplitude of the fMRI response can be determined for those voxels that pass a detection threshold. There is evidence from previous studies that the hemodynamic response may be time-locked to the stimulus within certain limits, less variable earlier in its evolution, and able to resolve information about relative hemodynamic timing. This information can be used to test hypotheses about the sequence and spatial distribution of neural activity. The method can be used to weight the earliest evolution of the hemodynamic response more heavily and decrease bias resulting from the hemodynamic response function. Additionally, the WLS method can control for varying response shapes across the brain and improve latency comparisons between brain regions. The WLS method was developed to study the properties of hemodynamic latencies, which may be increasingly important as event-related fMRI continues to be advanced.