When the t-value threshold is not reported in a paper, the advice is to identify the smallest peak they have reported with their uncorrected brain-wide p threshold, and select a t-value just below the value of that smallest peak. 

As I understand it, the "smallest peak" means the co-ordinate identified with the lowest t-value. E.g. if there were 3 peaks with t-values 4.5, 6.7, 9.8, the smallest peak here would be the one belonging to t-value 4.5, and the highest peak would be that of t-value 9.8. 

Is this understanding correct? 

I can't find any online explanations of what "smallest peak" actually means. 

Simon