Thanks for the replies Bruce and Doug. Do either of you know of any publications (or posters) that address the issue of editing effects in healthy controls?
Also do you know where I could find some examples of "bad" reconstructions? I have seen screenshots/images on the wiki, but I am looking for whole folders that contain all the files from the recon-all process. I was thinking outliers from a group mean in certain ratios (region A thickness/region B thickness, etc.) may provide some insight. -Andrew --------- I've looked at the actual editing of voxels (ie, erasing and deleting) vs putting control points in terms of their effect on a group analysis of aging. CPs have an appreciable effect. I could not find much of an effect of editing, probably because the edits rarely lined up across subject. I think the jury is still out. I think it is a hard t thing to evaluate because it can be population specific. doug On 05/04/2014 05:26 PM, Bruce Fischl wrote: Hi Andrew I'm not sure I have any useful advise for you. When/what to edit depends in part of your hypotheses, in part on how significant the error is, and in part on how easily fixed it is. You also have to be careful that you don't edit too much and significantly lower inter-rater (or intra-rater) reliability. Maybe Doug can comment as he has looked at the effects of editing vs. not editing in some large scale datasets and I believe hasn't found much of an effect. We've also done a number of inter-rater studies and not found any effect or rater (all unpublished). Sorry, as I said, I'm not sure how useful any of this is cheers Bruce On Fri, 2 May 2014, O'Shea,Andrew wrote: Hello FS experts, I have had a continuing curiosity regarding when/what to edit in FS cortical reconstructions. The wiki does a good job at showing how to deal with large scale problems, (i.e. invalid tal transformation, white matter not being recognized, large portions of the skull included in the surface) however I have yet to come across much that guides edits besides a "trained eye". How do we know when a problem is "big enough" that it needs attention? Are there any tool boxes/ quantitative guidelines that may help someone determine which scans need edits? I know QA tools outputs info on SNR/white matter intensity and I have looked for outliers here. However, I was wondering how else to automate/guide the decision process. Much of the discussion I have seen on this topic in manuscripts is either vague or nonexistent. Does anyone know of any manuscripts that have compared "raw" FS datasets to "selectively edited" datasets? What sort of effects do edits make in the grand scheme of a study? One could imagine problems with someone editing the data when they are knowledgable of the study's hypotheses and groups. Any additional info or guidance on this topic would be much appreciated. -Andrew _____________________________________________
_______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.