help
help > RE: How to describe the threshold of the result
Mar 30, 2017 08:03 PM | Alfonso Nieto-Castanon - Boston University
RE: How to describe the threshold of the result
Hi Xiuqin,
Your description is perfectly fine, the reviewer probably just wants you to be a bit more precise. Simply refer to "K.J. Friston, K.J. Worsley R.S.J. Frackowiak, J.C. Mazziotta, and A.C. Evans. Assessing the significance of focal activations using their spatial extent. Human Brain Mapping, 1:214-220, 1994" for additional information, and perhaps clarify that your voxel-level p<.001 threshold is the "height" or "cluster-defining" threshold, and that among the resulting suprathreshold clusters you are only reporting as significant those with cluster-extent FDR-corrected p-values below a p<.05 threshold (this is a multiple-comparisons correction acoss the entire brain that controls the chance of false discoveries among your reported clusters below the prescribed 5% rate).
Hope this helps
Alfonso
Originally posted by xiuqin jia:
Your description is perfectly fine, the reviewer probably just wants you to be a bit more precise. Simply refer to "K.J. Friston, K.J. Worsley R.S.J. Frackowiak, J.C. Mazziotta, and A.C. Evans. Assessing the significance of focal activations using their spatial extent. Human Brain Mapping, 1:214-220, 1994" for additional information, and perhaps clarify that your voxel-level p<.001 threshold is the "height" or "cluster-defining" threshold, and that among the resulting suprathreshold clusters you are only reporting as significant those with cluster-extent FDR-corrected p-values below a p<.05 threshold (this is a multiple-comparisons correction acoss the entire brain that controls the chance of false discoveries among your reported clusters below the prescribed 5% rate).
Hope this helps
Alfonso
Originally posted by xiuqin jia:
Hi all,
I have performed the seed-to-voxel FC analysis by CONN 17a recently. I have attached the result. I want to confirm whether the threhold I described was corrected and included all the informatin it should have. It goes as "Results were thresholded at voxel level p < 0.001 (uncorrected) and at cluster level p < 0.05 (FDR corrected)". If I'm wrong what it should be?
Based on the threshold I decribed above, the reviewer said that it should be described as "Results were thresholded at voxel level p<.001 and then corrected for whole-brain comparisons by False Discovery Rate (FDR) to q<.05" and asked "Was there then also a small cluster-size threshold applied, in addition to FDR? ". Actuall, it was the cluster-level FDR correcteion not the voxel-level correction. Does the cluster-level FDR correction should set a cluster-szie based on the voxel-level p < 0.001 uncorrected and how to do that in CONN? Does the default cluster-size of the p < 0.001 uncorrected at voxel level is zero? I am confused a little bit. Any comments are greatlly appreciated. Thank you.
Best, Xiuqin
I have performed the seed-to-voxel FC analysis by CONN 17a recently. I have attached the result. I want to confirm whether the threhold I described was corrected and included all the informatin it should have. It goes as "Results were thresholded at voxel level p < 0.001 (uncorrected) and at cluster level p < 0.05 (FDR corrected)". If I'm wrong what it should be?
Based on the threshold I decribed above, the reviewer said that it should be described as "Results were thresholded at voxel level p<.001 and then corrected for whole-brain comparisons by False Discovery Rate (FDR) to q<.05" and asked "Was there then also a small cluster-size threshold applied, in addition to FDR? ". Actuall, it was the cluster-level FDR correcteion not the voxel-level correction. Does the cluster-level FDR correction should set a cluster-szie based on the voxel-level p < 0.001 uncorrected and how to do that in CONN? Does the default cluster-size of the p < 0.001 uncorrected at voxel level is zero? I am confused a little bit. Any comments are greatlly appreciated. Thank you.
Best, Xiuqin
Threaded View
| Title | Author | Date |
|---|---|---|
| xiuqin jia | Mar 28, 2017 | |
| Alfonso Nieto-Castanon | Mar 30, 2017 | |
| xiuqin jia | Mar 31, 2017 | |