想请问一下各位老师关于做完TFCE矫正后结果的问题

审稿人给了这个意见:In terms of the VBM results, I found the description of 3-group differences a bit confusing and potentially misleading. The 3-group differences are at the cluster level, which includes the areas described as significant, but that assumption cannot be made. For example, the main contributor to cluster one is the right parahippocampal gyrus whereas the left parahippocampal gyrus is a discrete contributor to cluster two. Can the authors explain why they chose to describe only results at cluster level and not at peak level? Results at peak level would be important to discriminate regions of exact location of the peak effect and to decrease the chance of false positive rate. Can the authors also report the voxel-level peak statistics with coordinates, z values, and corrected p-values?

我想请教老师的有两点:

1.TFCE矫正后的P值不就是peak P-value吗?

2.他前面的意思是说无论团块包含了多少脑区只能讨论最显著的那个吗?

  1. TFCE校正后的P值就是peak-level的(或者是融合了peak和cluster extent信息的统计量),所以可能审稿人不太了解TFCE,也可能是因为你没有报告peak点的信息导致审稿人误解了。
  2. 我觉得审稿人不是这个意思。我猜测可能还是审稿人认为你做的是cluster-level的校正,所以如果是cluster-level的校正,是没有办法对peak点进行推断的,比如这个cluster里包含海马旁回,但并不能说海马旁回是显著的。cluster-level校正的结果只能说明这个cluster的extent是显著的。

谢谢老师解答,我还想问一下老师CAT12的TFCE校正做出来的结果只有P值和TFCE值据我自己的了解这种校正应该是没有z值的吧?

我不太记得CAT12是否会生成T值(FSL的randomise是会生成T值的),虽然TFCE是实际的统计量,但是没有太多信息,一般都不会报告的。如果CAT12没有生成T值,可以考虑根据SPM12的统计结果找到,因为在做TFCE校正之前,也需要做常规的peak-level的分析。T值和Z值可以相互转换,但是我不清楚报告Z值有什么好处,所以我觉得报告T值即可。

CAT12是没有生成T值,我再找找看,谢谢老师