Hi,
OK, that now explains things.
You can only get a value greater than one for the non-zero mean (-M) if the brain_lobe_mask is not binary (i.e. it contains values bigger than one). In that case you cannot use "-mul brain_lobe_mask" but would need to use "-mas brain_lobe_mask".
I would recommend sticking with the first case. You could fix the second case, by doing what I say above, which should make it more similar, but the first case is still the better processing option.
All the best,
Mark
On 24 Jan 2013, at 07:19, Ai Qing <[log in to unmask]<mailto:[log in to unmask]>> wrote:
Yes, fslstats masked_pve_1 -M -V
5.412853 900950 791850.562500 (I copied wrong here)
fslstats masked_pve_1 -M -V | awk '{print$1*$3}'
4.28617e+06 (This is correct)
The first fslstats line is correct.
But, the final result is correct. The two results are very different.
在 2013-1-24，下午3:12，Mark Jenkinson <[log in to unmask]<mailto:[log in to unmask]>> 写道：
Hi,
No, you misunderstand me. In both cases the second line just takes the numbers from the first line and multiplies the first and third numbers. Your mistake is that the numbers in the first line are identical in both cases. These numbers should be different. I think you have made a mistake copying into your email.
All the best,
Mark
Ai Qing <[log in to unmask]<mailto:[log in to unmask]>> wrote:
Dear Mark,
Sorry, I am a little confused. "...but should just be multiplying the first and third numbers together", here missed a "not"?
You mean that two methods are both correct, but the second should not calculate the volume just multiplying the first and third numbers? In my case, the best thrreshold is 0.5?
Thanks again!
Edwardai
在 2013-1-24，下午2:27，Mark Jenkinson <[log in to unmask]<mailto:[log in to unmask]>> 写道：
> Dear Ai Qing,
>
> You must have made a mistake here as the first fslstats line that you report is identical for the two cases, yet the second line is different but should just be multiplying the first and third numbers together. So check this all again carefully.
>
> All the best,
> Mark
>
>
> On 24 Jan 2013, at 04:41, Ai Qing <[log in to unmask]<mailto:[log in to unmask]>> wrote:
>
>> Dear Mark,
>>
>> I used two methods to calculate the brain lobe volume.
>> 1)fslmaths gm_pve -add wm_pve -mas brain_lobe_mask masked_pve; fslstats masked_pve -M -V | awk '{ print $1 * $3 }'
>>
>> I got the following result: fslstats masked_pve -M -V
>> 0.917858 963926 847200.562500
>>
>> fslstats masked_pve -M -V | awk '{print$1*$3}'
>> 777610
>>
>> 2) fslmaths gm_pve -add wm_pve GM_WM_pve; fslmaths GM_WM_pve -thr 0.5 -bin GM_WM_bin; fslmaths GM_WM_bin -mul brain_lobe_mask masked_pve_1
>>
>> The results was: fslstats masked_pve_1 -M -V
>> 0.917858 963926 847200.562500
>>
>> fslstats masked_pve_1 -M -V | awk '{print$1*$3}'
>> 4.28617e+06
>>
>> The two results are very different. I do not know what cause this.
>>
>> About the brain lobe atlas: I also refered to the Harvard anatomical lobe in FSL， but it is not suitable my study. So two doctors drew it on MNI152_brain_1mm in the ITK_SNAP for almost tow months. It's a very painful process.
>>
>> 在 2013-1-24，上午6:41，Mark Jenkinson <[log in to unmask]<mailto:[log in to unmask]>> 写道：
>>
>>> Hi,
>>>
>>> I'm not sure what you mean by absolute versus relative volume.
>>> The most accurate way to calculate the volume within a particular region, assuming you trust your brain_lobe_mask, is to:
>>> fslmaths gm_pve -add wm_pve -mas brain_lobe_mask masked_pve
>>> fslstats masked_pve -M -V | awk '{ print $1 * $3 }'
>>>
>>> However, the inaccuracies in the brain_lobe_mask are likely to dominate over the differences between thresholding the pve or not, but you might as well do it with the above commands.
>>>
>>> All the best,
>>> Mark
>>>
>>>
>>> On 23 Jan 2013, at 12:31, Ai Qing <[log in to unmask]<mailto:[log in to unmask]>> wrote:
>>>
>>>> Dear Mark,
>>>>
>>>> My understanding is that if I selected thr=0.5 to threshold the pve file, the volume will change. It will not be a absolute volume,but a relative volume. But if every patient has the same standard (I mean thr=0.5), it will not influence the statistic results. Do I understanding is right?
>>>>
>>>> Thanks again!
>>>>
>>>> 在 2013-1-8，下午6:18，Mark Jenkinson <[log in to unmask]<mailto:[log in to unmask]>> 写道：
>>>>
>>>>> Hi,
>>>>>
>>>>> This method sounds fine.
>>>>> One thing that isn't clear is what exactly you do with the PVEs.
>>>>> If you just want to exclude CSF then you could add up WM + GM as you do, or just use the CSF mask and invert it. Either way, you should threshold the mask as the last step. So adding up the WM and GM should be done first, and then the result should be thresholded and binarised (i.e. "-thr 0.5 -bin" in fslmaths). This then gives you a binary mask that you can use in step 5.
>>>>>
>>>>> All the best,
>>>>> Mark
>>>>>
>>>>>
>>>>>
>>>>> On 25 Dec 2012, at 02:21, Ai Qing <[log in to unmask]<mailto:[log in to unmask]>> wrote:
>>>>>
>>>>>> Dear FSL experts:
>>>>>>
>>>>>> Merry Christmas!
>>>>>>
>>>>>> I do not know whether the following method of caliculating brain lobe volume is reasonable or not. I hope to get some valuable advice from you.
>>>>>>
>>>>>> Modality: MPRAGE
>>>>>>
>>>>>> 1) Draw a brain anatomical region mask on MNI152_1mm ( or using juelich atlas), I call Brain_template
>>>>>> 2) Transform brain_anatomical_mask to native MPRAGE space using invwarp.
>>>>>> 3) Segement MPRAGE to GM/WM using FAST, and get pve file( this is the most accurate way to calculate volume according to the UserGuide).
>>>>>> 4) Combing GM and WM mask together (brain_lobe_mask) using fslmaths (-add)
>>>>>> 5) fslmaths Brain_template -mul brain_lobe_mask to get brain_lobe of different anatomical region.
>>>>>> 6) using fslstas to calculate volume of different brain lobe.
>>>>>>
>>>>>> I really do not know whether the method is reasonable or not.
>>>>>>
>>>>>> Thanks again.
>>>>>>
>>>>>> Merry Christmas!!
>>>
>
