dke-questions > Strage pxel blocks on fitting Kmeans maps and parts of the image are missing
Showing 1-4 of 4 posts
Jan 20, 2022 12:01 PM | ruicheng ba
Strage pxel blocks on fitting Kmeans maps and parts of the image are missing
I input dwi images with b=0,0.8,1.5 and 2.5, where b0 has one
direction and the other b values have 18 directions. There are
discontinuous squares on the Kmean maps of the output result. May I
ask why this is caused? I checked the original image, and there
were no artifacts of similar shapes.
In addition, I found that part of images on Kmean Map were missing. I attached the comparison images. Attachment 1 is the Kmean map, and Attachment 2 is the B0 image of the input data.
In addition, I found that part of images on Kmean Map were missing. I attached the comparison images. Attachment 1 is the Kmean map, and Attachment 2 is the B0 image of the input data.
Mar 28, 2022 02:03 PM | Emilie McKinnon - MUSC
RE: Strage pxel blocks on fitting Kmeans maps and parts of the image are missing
Hi!
Check out the masking feature. By default, DKE uses a threshold value of 50 in the B-0 value to create a brain mask. Just by glancing at your image it looks like this is animal data and so your threshold might have to be adjusted?
Just so you know, we are phasing out support for DKE and have moved to a more modern software that will be monitored more closely. The documentation can be found here: https://pydesigner.readthedocs.io/en/latest/
I apologize for the delay!
Best,
Emilie
Check out the masking feature. By default, DKE uses a threshold value of 50 in the B-0 value to create a brain mask. Just by glancing at your image it looks like this is animal data and so your threshold might have to be adjusted?
Just so you know, we are phasing out support for DKE and have moved to a more modern software that will be monitored more closely. The documentation can be found here: https://pydesigner.readthedocs.io/en/latest/
I apologize for the delay!
Best,
Emilie
Mar 29, 2022 08:03 AM | ruicheng ba
RE: Strage pxel blocks on fitting Kmeans maps and parts of the image are missing
Thank you for your reply!
I have tried the directional fitting instead of tensor fitting and I found these pixels disappear and the fitting results became coherent. So what's the difference between these two fitting methods? If I acquired data in >15 directions, which methods could be more accurate? By the way, is there any requirements on adjusting the B-0 threshold?
Best,
Ruicheng
I have tried the directional fitting instead of tensor fitting and I found these pixels disappear and the fitting results became coherent. So what's the difference between these two fitting methods? If I acquired data in >15 directions, which methods could be more accurate? By the way, is there any requirements on adjusting the B-0 threshold?
Best,
Ruicheng
Mar 29, 2022 08:03 AM | ruicheng ba
RE: Strage pxel blocks on fitting Kmeans maps and parts of the image are missing
Originally posted by Emilie McKinnon:
Thank you for your reply! I have tried the directional fitting instead of tensor fitting and I found these pixels disappear and the fitting results became coherent. So what's the difference between these two fitting methods? If I acquired data in >15 directions, which methods could be more accurate? By the way, is there any requirements on adjusting the B-0 threshold?
Best,
Ruicheng
Hi!
Check out the masking feature. By default, DKE uses a threshold value of 50 in the B-0 value to create a brain mask. Just by glancing at your image it looks like this is animal data and so your threshold might have to be adjusted?
Just so you know, we are phasing out support for DKE and have moved to a more modern software that will be monitored more closely. The documentation can be found here: https://pydesigner.readthedocs.io/en/latest/
I apologize for the delay!
Best,
Emilie
Check out the masking feature. By default, DKE uses a threshold value of 50 in the B-0 value to create a brain mask. Just by glancing at your image it looks like this is animal data and so your threshold might have to be adjusted?
Just so you know, we are phasing out support for DKE and have moved to a more modern software that will be monitored more closely. The documentation can be found here: https://pydesigner.readthedocs.io/en/latest/
I apologize for the delay!
Best,
Emilie
Thank you for your reply! I have tried the directional fitting instead of tensor fitting and I found these pixels disappear and the fitting results became coherent. So what's the difference between these two fitting methods? If I acquired data in >15 directions, which methods could be more accurate? By the way, is there any requirements on adjusting the B-0 threshold?
Best,
Ruicheng