patch2self denoising attenuated b=0 images
Hi, I was trying out the new patch2self denoising on a 4-shell (b=1k s/mm^2 up to 4k s/mm^2) diffusion data (total of 94 directions plus 4 b=0 images). First, I noticed that it attenuated the b=0 images (all four of them) with respect to the rest of the image series. This in turn reduced map-mri ZD maps by about 10% to 20%, which was expected b/c (E(q)) essentially got "smaller". Did anybody have any experience with this? Any suggestion why it happened and how to fix it? best, -Tugan
Hi Tugan, Thank you for your question! This effect that you are seeing could be happening due to one of 2 reasons: 1. The b0_threshold parameter used in DIPY is set to 50. If the value of your b0 volumes is greater than 50 in your bval, it would be assumed as a DWI volume, which may cause this effect. 2. You need to set the parameters shift_intensity=True and clip_negative_vals=False within Patch2Self. If setting both of these above points don't work for you, you can also skip denoising b0 volumes by using the parameter: b0_denoising=False If possible, I can also take a look at your dataset and help you with it! Regards, Shreyas ________________________________ From: lmuftuler--- via DIPY <dipy@python.org> Sent: Wednesday, March 31, 2021 5:10 PM To: dipy@python.org <dipy@python.org> Subject: [External] [DIPY] patch2self denoising attenuated b=0 images This message was sent from a non-IU address. Please exercise caution when clicking links or opening attachments from external sources. ------- Hi, I was trying out the new patch2self denoising on a 4-shell (b=1k s/mm^2 up to 4k s/mm^2) diffusion data (total of 94 directions plus 4 b=0 images). First, I noticed that it attenuated the b=0 images (all four of them) with respect to the rest of the image series. This in turn reduced map-mri ZD maps by about 10% to 20%, which was expected b/c (E(q)) essentially got "smaller". Did anybody have any experience with this? Any suggestion why it happened and how to fix it? best, -Tugan _______________________________________________ DIPY mailing list -- dipy@python.org To unsubscribe send an email to dipy-leave@python.org https://mail.python.org/mailman3/lists/dipy.python.org/ Member address: shfadn@iu.edu
Hi Shreyas, Thank you very much for the suggestions. Let me try these options and see if one of them fixes the problem. It is not #1 (b0 threshold), though. I'll keep you posted when I get a chance to test these. best, -Tugan ---- ________________________________ From: Fadnavis, Shreyas Sanjeev <shfadn@iu.edu> Sent: Wednesday, March 31, 2021 04:46 PM To: dipy@python.org <dipy@python.org>; Muftuler, L. Tugan <lmuftuler@mcw.edu> Subject: Re: [External] [DIPY] patch2self denoising attenuated b=0 images ATTENTION: This email originated from a sender outside of MCW. Use caution when clicking on links or opening attachments. ________________________________ Hi Tugan, Thank you for your question! This effect that you are seeing could be happening due to one of 2 reasons: 1. The b0_threshold parameter used in DIPY is set to 50. If the value of your b0 volumes is greater than 50 in your bval, it would be assumed as a DWI volume, which may cause this effect. 2. You need to set the parameters shift_intensity=True and clip_negative_vals=False within Patch2Self. If setting both of these above points don't work for you, you can also skip denoising b0 volumes by using the parameter: b0_denoising=False If possible, I can also take a look at your dataset and help you with it! Regards, Shreyas ________________________________ From: lmuftuler--- via DIPY <dipy@python.org> Sent: Wednesday, March 31, 2021 5:10 PM To: dipy@python.org <dipy@python.org> Subject: [External] [DIPY] patch2self denoising attenuated b=0 images This message was sent from a non-IU address. Please exercise caution when clicking links or opening attachments from external sources. ------- Hi, I was trying out the new patch2self denoising on a 4-shell (b=1k s/mm^2 up to 4k s/mm^2) diffusion data (total of 94 directions plus 4 b=0 images). First, I noticed that it attenuated the b=0 images (all four of them) with respect to the rest of the image series. This in turn reduced map-mri ZD maps by about 10% to 20%, which was expected b/c (E(q)) essentially got "smaller". Did anybody have any experience with this? Any suggestion why it happened and how to fix it? best, -Tugan _______________________________________________ DIPY mailing list -- dipy@python.org To unsubscribe send an email to dipy-leave@python.org https://mail.python.org/mailman3/lists/dipy.python.org/<https://urldefense.com/v3/__https://mail.python.org/mailman3/lists/dipy.python.org/__;!!H8mHWRdzp34!pg5c4eEHbhaCEU6gj-z5fTexe1OtVbJ4UgKLAisIFcaGbsoQydJ3dmgrx7sckqSk$> Member address: shfadn@iu.edu
Hi Shreyas, Your suggestion #2 worked. I checked signal evolution in several WM tracts and the signal changes are in accord with the original diffusion data. I am impressed by how well the images are denoised without losing details. It would be great if you add these suggestions to the DIPY manual for other users who might encounter the same issue. Thanks again for your help. best regards, - Tugan - ________________________________ From: Fadnavis, Shreyas Sanjeev <shfadn@iu.edu> Sent: Wednesday, March 31, 2021 04:46 PM To: dipy@python.org <dipy@python.org>; Muftuler, L. Tugan <lmuftuler@mcw.edu> Subject: Re: [External] [DIPY] patch2self denoising attenuated b=0 images ATTENTION: This email originated from a sender outside of MCW. Use caution when clicking on links or opening attachments. ________________________________ Hi Tugan, Thank you for your question! This effect that you are seeing could be happening due to one of 2 reasons: 1. The b0_threshold parameter used in DIPY is set to 50. If the value of your b0 volumes is greater than 50 in your bval, it would be assumed as a DWI volume, which may cause this effect. 2. You need to set the parameters shift_intensity=True and clip_negative_vals=False within Patch2Self. If setting both of these above points don't work for you, you can also skip denoising b0 volumes by using the parameter: b0_denoising=False If possible, I can also take a look at your dataset and help you with it! Regards, Shreyas ________________________________ From: lmuftuler--- via DIPY <dipy@python.org> Sent: Wednesday, March 31, 2021 5:10 PM To: dipy@python.org <dipy@python.org> Subject: [External] [DIPY] patch2self denoising attenuated b=0 images This message was sent from a non-IU address. Please exercise caution when clicking links or opening attachments from external sources. ------- Hi, I was trying out the new patch2self denoising on a 4-shell (b=1k s/mm^2 up to 4k s/mm^2) diffusion data (total of 94 directions plus 4 b=0 images). First, I noticed that it attenuated the b=0 images (all four of them) with respect to the rest of the image series. This in turn reduced map-mri ZD maps by about 10% to 20%, which was expected b/c (E(q)) essentially got "smaller". Did anybody have any experience with this? Any suggestion why it happened and how to fix it? best, -Tugan _______________________________________________ DIPY mailing list -- dipy@python.org To unsubscribe send an email to dipy-leave@python.org https://mail.python.org/mailman3/lists/dipy.python.org/<https://urldefense.com/v3/__https://mail.python.org/mailman3/lists/dipy.python.org/__;!!H8mHWRdzp34!pg5c4eEHbhaCEU6gj-z5fTexe1OtVbJ4UgKLAisIFcaGbsoQydJ3dmgrx7sckqSk$> Member address: shfadn@iu.edu
Hi Tugan, Thank you for your question! This effect that you are seeing could be happening due to one of 2 reasons: 1. The b0_threshold parameter used in DIPY is set to 50. If the value of your b0 volumes is greater than 50 in your bval, it would be assumed as a DWI volume, which may cause this effect. 2. You need to set the parameters shift_intensity=True and clip_negative_vals=False within Patch2Self. If setting both of these above points don't work for you, you can also skip denoising b0 volumes by using the parameter: b0_denoising=False If possible, I can also take a look at your dataset and help you with it! Regards, Shreyas ________________________________ From: lmuftuler--- via DIPY <dipy@python.org> Sent: Wednesday, March 31, 2021 5:10 PM To: dipy@python.org <dipy@python.org> Subject: [External] [DIPY] patch2self denoising attenuated b=0 images This message was sent from a non-IU address. Please exercise caution when clicking links or opening attachments from external sources. ------- Hi, I was trying out the new patch2self denoising on a 4-shell (b=1k s/mm^2 up to 4k s/mm^2) diffusion data (total of 94 directions plus 4 b=0 images). First, I noticed that it attenuated the b=0 images (all four of them) with respect to the rest of the image series. This in turn reduced map-mri ZD maps by about 10% to 20%, which was expected b/c (E(q)) essentially got "smaller". Did anybody have any experience with this? Any suggestion why it happened and how to fix it? best, -Tugan _______________________________________________ DIPY mailing list -- dipy@python.org To unsubscribe send an email to dipy-leave@python.org https://mail.python.org/mailman3/lists/dipy.python.org/ Member address: shfadn@iu.edu
Hi Shreyas, Thank you very much for the suggestions. Let me try these options and see if one of them fixes the problem. It is not #1 (b0 threshold), though. I'll keep you posted when I get a chance to test these. best, -Tugan ---- ________________________________ From: Fadnavis, Shreyas Sanjeev <shfadn@iu.edu> Sent: Wednesday, March 31, 2021 04:46 PM To: dipy@python.org <dipy@python.org>; Muftuler, L. Tugan <lmuftuler@mcw.edu> Subject: Re: [External] [DIPY] patch2self denoising attenuated b=0 images ATTENTION: This email originated from a sender outside of MCW. Use caution when clicking on links or opening attachments. ________________________________ Hi Tugan, Thank you for your question! This effect that you are seeing could be happening due to one of 2 reasons: 1. The b0_threshold parameter used in DIPY is set to 50. If the value of your b0 volumes is greater than 50 in your bval, it would be assumed as a DWI volume, which may cause this effect. 2. You need to set the parameters shift_intensity=True and clip_negative_vals=False within Patch2Self. If setting both of these above points don't work for you, you can also skip denoising b0 volumes by using the parameter: b0_denoising=False If possible, I can also take a look at your dataset and help you with it! Regards, Shreyas ________________________________ From: lmuftuler--- via DIPY <dipy@python.org> Sent: Wednesday, March 31, 2021 5:10 PM To: dipy@python.org <dipy@python.org> Subject: [External] [DIPY] patch2self denoising attenuated b=0 images This message was sent from a non-IU address. Please exercise caution when clicking links or opening attachments from external sources. ------- Hi, I was trying out the new patch2self denoising on a 4-shell (b=1k s/mm^2 up to 4k s/mm^2) diffusion data (total of 94 directions plus 4 b=0 images). First, I noticed that it attenuated the b=0 images (all four of them) with respect to the rest of the image series. This in turn reduced map-mri ZD maps by about 10% to 20%, which was expected b/c (E(q)) essentially got "smaller". Did anybody have any experience with this? Any suggestion why it happened and how to fix it? best, -Tugan _______________________________________________ DIPY mailing list -- dipy@python.org To unsubscribe send an email to dipy-leave@python.org https://mail.python.org/mailman3/lists/dipy.python.org/<https://urldefense.com/v3/__https://mail.python.org/mailman3/lists/dipy.python.org/__;!!H8mHWRdzp34!pg5c4eEHbhaCEU6gj-z5fTexe1OtVbJ4UgKLAisIFcaGbsoQydJ3dmgrx7sckqSk$> Member address: shfadn@iu.edu
Hi Shreyas, Your suggestion #2 worked. I checked signal evolution in several WM tracts and the signal changes are in accord with the original diffusion data. I am impressed by how well the images are denoised without losing details. It would be great if you add these suggestions to the DIPY manual for other users who might encounter the same issue. Thanks again for your help. best regards, - Tugan - ________________________________ From: Fadnavis, Shreyas Sanjeev <shfadn@iu.edu> Sent: Wednesday, March 31, 2021 04:46 PM To: dipy@python.org <dipy@python.org>; Muftuler, L. Tugan <lmuftuler@mcw.edu> Subject: Re: [External] [DIPY] patch2self denoising attenuated b=0 images ATTENTION: This email originated from a sender outside of MCW. Use caution when clicking on links or opening attachments. ________________________________ Hi Tugan, Thank you for your question! This effect that you are seeing could be happening due to one of 2 reasons: 1. The b0_threshold parameter used in DIPY is set to 50. If the value of your b0 volumes is greater than 50 in your bval, it would be assumed as a DWI volume, which may cause this effect. 2. You need to set the parameters shift_intensity=True and clip_negative_vals=False within Patch2Self. If setting both of these above points don't work for you, you can also skip denoising b0 volumes by using the parameter: b0_denoising=False If possible, I can also take a look at your dataset and help you with it! Regards, Shreyas ________________________________ From: lmuftuler--- via DIPY <dipy@python.org> Sent: Wednesday, March 31, 2021 5:10 PM To: dipy@python.org <dipy@python.org> Subject: [External] [DIPY] patch2self denoising attenuated b=0 images This message was sent from a non-IU address. Please exercise caution when clicking links or opening attachments from external sources. ------- Hi, I was trying out the new patch2self denoising on a 4-shell (b=1k s/mm^2 up to 4k s/mm^2) diffusion data (total of 94 directions plus 4 b=0 images). First, I noticed that it attenuated the b=0 images (all four of them) with respect to the rest of the image series. This in turn reduced map-mri ZD maps by about 10% to 20%, which was expected b/c (E(q)) essentially got "smaller". Did anybody have any experience with this? Any suggestion why it happened and how to fix it? best, -Tugan _______________________________________________ DIPY mailing list -- dipy@python.org To unsubscribe send an email to dipy-leave@python.org https://mail.python.org/mailman3/lists/dipy.python.org/<https://urldefense.com/v3/__https://mail.python.org/mailman3/lists/dipy.python.org/__;!!H8mHWRdzp34!pg5c4eEHbhaCEU6gj-z5fTexe1OtVbJ4UgKLAisIFcaGbsoQydJ3dmgrx7sckqSk$> Member address: shfadn@iu.edu
participants (4)
-
Fadnavis, Shreyas Sanjeev -
lmuftuler＠mcw.edu -
Muftuler, L. Tugan
-
Shreyas Fadnavis