请教关于平滑核的问题

Submitted by 七色鱼 on
请教老师,为什么改了平滑核ALFF会有很大改变,原来是increase的会变成decrease,这样合理吗,MCI与正常人相比,后扣带ALFFincrease可能吗,急求老师指教。

Qing Zhao

Mon, 07/16/2012 - 18:38

Smooth is kind of average of one voxel and its nearby neighbors. If a voxel has some nearby neighbors which has very different value, after smooth the value on this voxel will changed, thus changed the result probably.
Different FWHM define different size of the area to be averaged. Although which is the best FWHM is very difficult to decide, I guess if the area you interested located near boundaries of different tissues, you'd better to not use large FWHM.    
 

Thank you for the quick answer.I compare the MCI group with the control group,but the result is opposite with previous study.I have checked every step,no mistakes is found.Could this be possible?I tried falff,too.And the result is opposite with the result using alff.Can you tell me what I should do?I'll appreciate  your answer  very much.

Sorry for not familiar with MCI. You can wait for other one who can answer you about this.

But I suggest to make clear that previous study is in ReHo, ALFF or functional connectivity.
Different measurement reflect different things so it may be "opposite".
If exactly same things (including reprocessing steps, subject's age, gender, race and so on) in your results are opposite with previous ones, the possible explain is that fMRI signal of this disease may have a very unstable behavior cross different subjects.

But you need to check all the steps and all subject's status very carefully.
 


在同一个体素来说,几乎不可能出现这种情况:原来是A组>B组,平滑之后B组>A组。我推测,你说的“同一个部位”,并非真正同一个部位。MCI或AD在PCC(一般来说是BA 23附近)葡萄糖代谢减低、ALFF减低、ReHo减低,已经被许多研究报道。可否提供你的结果的peak坐标?在BA哪个区?

peak点坐标(-5,-48,27),差不多就是在23区。而且亮的地方全都是MCI>HC,不知道怎么可以发图给您。我筛选被试很严格,MMSE和N5等测验得分都卡得很严格、配准、头动不好的都很删掉了,也试着随机删除过几个被试,但是结果一直都不好。我也试了malff和falff,falff结果和alff好多都是相反的,这个我也不太明白。alff\malff\falff到底选哪一个更好,关于结果不好的问题我能做些什么呢,请臧老师指教。
Forums