it’s not a t-value. These values are the temporal extent (in samples) of that vertex in the cluster. A cluster encompasses different vertices at different time points.
Generally, there is no “statistically significant region” as a result from a cluster-based permutation test. I know the fact that you have clusters and every cluste as a p-value associated with it is really inviting to interpretations as “in these areas there is a significant difference and in other areas there is not”, but this would be an incorrect interpretation of the test. See also: https://onlinelibrary.wiley.com/doi/10.1111/psyp.13335