 |
 |
Dear Paul, I have tested the FAST with bias-field correction turned off, the results were better but still not right. The FAST works well on the unprocessed image. I will try your suggestions: run FAST on the unprocessed, brain-extracted image, and then transform the outputs of FAST to a different space if needed. Thanks for your help. Best regards, Liangqiong ________________________________ 发件人: FSL - FMRIB's Software Library <[log in to unmask]> 代表 paul mccarthy <[log in to unmask]> 发送时间: 2019年7月19日 21:46 收件人: [log in to unmask] <[log in to unmask]> 主题: Re: [FSL] 答复: [FSL] 答复: [FSL] [FSL] FAST Segmentation problem with all the brain regions are classified as GM Hi Qiong, Given that you have already performed bias-field correction, did you try running FAST with its bias-field correction turned off, as Taylor suggested? It is also uncommon to run FAST on an image which has been registered to a different space - you will generally get better results if you run FAST on the unprocessed, brain-extracted image (again, suggested by Taylor). You can then transform the outputs of FAST to a different space if needed. Paul On 19/07/2019, QU Liangqiong <[log in to unmask]> wrote: > Dear Paul, > > > Still not work if I skip the normalization step. > > I have tried ANTs, it seems that it works well on my data. > > > Best regards, > > Liangqiong > > > ________________________________ > 发件人: FSL - FMRIB's Software Library <[log in to unmask]> 代表 paul mccarthy > <[log in to unmask]> > 发送时间: 2019年7月19日 16:34 > 收件人: [log in to unmask] <[log in to unmask]> > 主题: Re: [FSL] 答复: [FSL] [FSL] FAST Segmentation problem with all the brain > regions are classified as GM > > Hi Qiong, > > Does FAST produce better results if you skip the normalisation step? > > Cheers, > > Paul > > On 16/07/2019, QU Liangqiong <[log in to unmask]> wrote: >> Hello, >> >> Thank you very much for your illustration. >> >> The pre-processing steps are: >> >> 1. linear registration to MNI space >> >> 2. inhomogeneity correction >> >> 3. skull-stripping >> >> 4. normalization to [0,1] >> >> >> The results are better when adding -N option, but still not quite >> correct. >> >> See the attachment for an example with the -N option. >> >> [cid:42818c51-0c4d-40f6-916b-c516b8451d29] >> >> >> FAST work well on the original image. >> >> It seems that the image pre-processing steps cause the problem. >> >> >> Thanks very much for your help. >> >> >> Best regards, >> >> Liangqiong >> >> ________________________________ >> 发件人: FSL - FMRIB's Software Library <[log in to unmask]> 代表 Taylor >> Hanayik >> <[log in to unmask]> >> 发送时间: 2019年7月15日 17:41:41 >> 收件人: [log in to unmask] >> 主题: Re: [FSL] [FSL] FAST Segmentation problem with all the brain regions >> are >> classified as GM >> >> Hi, >> >> I’ve had a look at the images you uploaded and they seem to have been >> preprocessed in some way. >> >> Could you describe what processing has already been done to the images >> prior >> to FAST? >> >> FAST is failing at the bias correction stage. Have you previously bias >> corrected this data before running FAST? If so, you can turn off FAST’s >> bias >> correction stage by using the -N option. >> >> Also, the data has been scaled to a 0-1 range. Was that step part of some >> preprocessing before FAST? >> >> Could you try running the unprocessed (brain extracted) T1 through FAST >> to >> see if things work ok? >> >> >> >> >> Cheers, >> >> Taylor Hanayik >> >> >> Taylor Hanayik PhD >> Analysis Research Software Engineer >> FMRIB, John Radcliffe Hospital >> University of Oxford >> [log in to unmask]<mailto:[log in to unmask]> >> >> >> >> >> >> On 9 Jul 2019, at 19:51, QU Liangqiong >> <[log in to unmask]<mailto:[log in to unmask]>> >> wrote: >> >> Hello, >> >> I have updated two samples of my T1 files (the input to FAST) to >> https://oxfile.ox.ac.uk/oxfile/work/extBox?id=754626B883867C03BB, with >> name >> S1_7t.nii and S7_3t.nii. >> >> Thank you very much for your help. >> >> Best regards, >> Liangqiong >> ________________________________ >> 发件人: FSL - FMRIB's Software Library >> <[log in to unmask]<mailto:[log in to unmask]>> 代表 Taylor Hanayik >> <[log in to unmask]<mailto:[log in to unmask]>> >> 发送时间: 2019年7月10日 2:40:37 >> 收件人: [log in to unmask]<mailto:[log in to unmask]> >> 主题: Re: [FSL] FAST Segmentation problem with all the brain regions are >> classified as GM >> >> Hi, >> >> Happy to look into it for you. Could you upload a T1 file (the input that >> you give to FAST)? We’ll have a look and test on our end. >> >> You can upload to Oxford’s file sharing service here: >> >> https://oxfile.ox.ac.uk/oxfile/work/extBox?id=754626B883867C03BB >> >> >> Cheers, >> >> Taylor Hanayik >> >> >> Taylor Hanayik PhD >> Analysis Research Software Engineer >> FMRIB, John Radcliffe Hospital >> University of Oxford >> [log in to unmask]<mailto:[log in to unmask]> >> >> >> >> >> >> On 7 Jul 2019, at 01:41, Liangqiong Qu >> <[log in to unmask]<mailto:[log in to unmask]>> >> wrote: >> >> Hello! >> >> I want to use “FAST Segmentation” to segment GM/WM/CSF from a >> skull-stripped >> 3T T1-weighted MR image. But it classifies all the brain regions as GM. >> I >> have tuned the parameters, but it still not work. >> >> >> I used the command like “/usr/local/fsl/bin/fast -t 1 -n 3 -H 0.1 -I 4 -l >> 20.0 -o $Output_dir $Input_dir ” >> >> >> I have also tested the same command on some other types of 3T T1-weighted >> MR >> image, the result seems good. >> >> >> I want to ask: >> >> Is there any requirement of the input of T1-weighted MR image? Why it >> works >> well on some types of T1-weighted MR image work well, but collapses on >> other >> types of T1-weighted images. >> >> Could you give me some advice on this? >> >> I am looking forward for your feedback. >> >> Best regards, >> Qiong >> >> ######################################################################## >> >> To unsubscribe from the FSL list, click the following link: >> https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1 >> >> >> ________________________________ >> To unsubscribe from the FSL list, click the following link: >> https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1 >> >> ________________________________ >> To unsubscribe from the FSL list, click the following link: >> https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1 >> >> >> ________________________________ >> >> To unsubscribe from the FSL list, click the following link: >> https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1 >> >> ######################################################################## >> >> To unsubscribe from the FSL list, click the following link: >> https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1 >> > > ######################################################################## > > To unsubscribe from the FSL list, click the following link: > https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1 > > ######################################################################## > > To unsubscribe from the FSL list, click the following link: > https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1 > ######################################################################## To unsubscribe from the FSL list, click the following link: https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1 ######################################################################## To unsubscribe from the FSL list, click the following link: https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1