 |
 |
Dear Faezeh, On 31 May 2017, at 19:20, Faezeh Vedaei wrote: > Dear Dr. Gaser > Thank you for your response. I do not have any idea why everything is > weird for this data among many other flair data I have done LST > segmentation. I tried applying LGA algorithm again by using several > threshold between 0 and 1. The outputs show nothing yet. In case I > would like to ask you this, the only difference between this data with > other our data is that in this Flair data some parts of the bottom of > the brain is cut, but I thought it may not affect our result. So, is > this possible this is the reason this segmentation could not be done > well? Maybe you just found the reason for the failed segmentation. Brains without full FOV are always a mess to process and this will usually fail. Best, Christian > RegardsFaezeh > > On Wednesday, May 31, 2017 11:48 AM, Christian Gaser > <[log in to unmask]> wrote: > > > Dear Faezeh,the LGA approach can be adapted to your data using > different thresholds. You can define a row of different values between > 0..1 to get lesion results for these > thresholds.Best,ChristianChristian Gaser, Ph.D. > Professor of Computational Neuroscience/Neuroimaging > Biomagnetic Center > Structural Brain Mapping Group > Department of Neurology > Jena University Hospital > Am Klinikum 1, D-07747 Jena, Germany > Tel: ++49-3641-9325778 Fax: ++49-3641-9325772 > e-mail: [log in to unmask] > http://www.neuro.uni-jena.deOn 31 May 2017, at 17:46, Faezeh Vedaei > wrote: > Dear Dr. Gaser > Thank you for your response. Actually, I tried using LGA algorithm as > well, but the result of segmentation does not show anything. Indeed, > the flair map (lesion part) is blank in this output, whereas we know > that there are parts of hyperintensity lesion in white matter. So, > what could be the problem? Do you think there is other way I can fix > this problem? I really appreciate your help and time to us in this > regard. > Bests,Faezeh > > On Wednesday, May 31, 2017 10:08 AM, Christian Gaser > <[log in to unmask]> wrote: > > > Dear Faezeh, > > the LPA approach is based on the segmentation of the FLAIR images > only, which sometimes results in less accurate results depending on > the lesions in the image. This approach is mainly thought if you don't > have an additional T1 image and/or it is difficult to find a stable > threshold for the LGA approach. The multimodal segmentation of > T1+FLAIR using the LGA approach might be the more appropriate approach > if LPA fails. Fortunately, this only happens for one data set even if > this is disappointing anyway. The LPA approach does not allow to > control the segmentation using a threshold... > > Best, > > Christian > > On Mon, 22 May 2017 20:30:08 +0100, Faeze Vedaei <[log in to unmask]> > wrote: > >> Hi SPMers >> >> I have used LST toolbox on SPM (LPA algorithm) to do Flair >> segmentation to hyperintensity regions many times. The issue is that >> for just one of my subjects which Fair data seems OK, the results of >> LPA does not make sense. Indeed, Flair map as the result of Flair >> segementation is so exaggerated by enlarged hyperintensity areas. I >> do not have any idea why this happened to this data and how I can fix >> that to a real segmented result. Here, I provided two screen shots of >> original Flair and Flair map (result of segmentation) superimposed on >> Flair data, you can see and compare how hyprintensity areas are in >> different size is these two images. Any help to solve my problem is >> appreciated and please let me know if you need some more information. >> Thank you so much. >> >> Regards, >> Faezeh Vedaei >>