Re: [Freesurfer] SAMSEG "lesion" (Illegal instruction) ~ running VM_67

Hi Randy, Your run_samseg syntax is correct and yes, the "lesion" version of SAMSEG requires Tensorflow. The "illegal instruction" error is most probably related to an incompatible version of Tensorflow with your CPU instruction set. You should be able to replicate the error by typing: 1.

[Freesurfer] R: SAMSEG "lesion" (Illegal instruction) ~ running VM_67

omeone else on the list can solve your issue. Stefano Da: daedalu...@aol.com Inviato: lunedì 28 febbraio 2022 14:52 A: Cerri, Stefano ; freesurfer@nmr.mgh.harvard.edu Oggetto: Re: [Freesurfer] SAMSEG "lesion" (Illegal instruction) ~ running VM_67 Exte

Re: [Freesurfer] SAMSEG Lesion images not written

Hi Randy, "seg.mgz" is the "hard segmentation", with lesion labeled as 99. You have all the volumes in "samseg.stats". If you want segmentation maps (values between 0 and 1), you should add the flag --save-posteriors to run_samseg. They will be saved in the samseg output directory under the "

Re: [Freesurfer] SAMSEG (Lesion Mask Pattern) ... ah, [0 0]

Hi Randy, The default option is 0 for each input scan as no assumption on the MRI contrast and order of the input scans is made. We strongly recommend users to use the value of 1 for both T2w and FLAIR scans and 0 for T1w scans. The mask is reducing false-positive lesions by applying intensity

[Freesurfer] Incorporating samseg segmentation into the recon-all pipeline

Hi Julian, You can use the command "samseg" with the --recon flag. See the helper function for more details (--help) Stefano External Email - Use Caution /Hello everyone,/ /We have been utilizing freesurfer for our brain morphometry and processing pipelines, and have recently tested the sa

[Freesurfer] Large number of false negative MS lesions in SAMSEG

Hi Martin, Can you try to remove the lesion intensity constraint on the T2 data (i.e., by modifying your command with --lesion-mask-pattern 0 1 0) for a couple of cases and let us know if it helps? If not, would you be able to share with us one case where it fails? FYI: you can save a lesion

[Freesurfer] R: Large number of false negative MS lesions in SAMSEG

upload it? Thanks, Martin On Sun, Aug 14, 2022 at 10:52 PM Cerri, Stefano mailto:sce...@mgh.harvard.edu>> wrote: Hi Martin, Can you try to remove the lesion intensity constraint on the T2 data (i.e., by modifying your command with --lesion-mask-pattern 0 1 0) for a couple of cases a

[Freesurfer] samseg lesions labels and values

Hi Dilek, Yes, "--save-posteriors Lesions" will save a probability map with path *yourOutputDir*/posteriors/Lesions.mgz 1 and 2) Lesions are labeled as 99 in seg.mgz. A value of 0.04 in the stats file indicate that the lesion volume is 0.04 mm^3, so it makes sense that seg.mgz has no voxels la

[Freesurfer] R: SAMSEG estimation error

://surfer.nmr.mgh.harvard.edu/fswiki/BuildGuide Hope it helps, Stefano Da: giulio siracusano Inviato: sabato 27 agosto 2022 04:22 A: freesurfer@nmr.mgh.harvard.edu Cc: Cerri, Stefano Oggetto: Re: [Freesurfer] SAMSEG estimation error External Email - Use Caution Hi

[Freesurfer] R: SAMSEG estimation error

y the way, we have fixed in version 7.3 the seed to obtain reproducible results. Are you sure you're running these commands on the same machine? Stefano Da: giulio siracusano Inviato: sabato 27 agosto 2022 14:15 A: Cerri, Stefano Cc: giulio siracusano ;

Re: [Freesurfer] samseg inputs

Hi Dilek, I've actually run SAMSEG-lesion on some gadolinium-enhanced scans in the paper. Look at the top row of Fig. 7 ("T1c") and at the "MSseg" dataset. I did not check the performance of the method on such data. We also don't distinguish between enhanced and non-enhanced lesions. I would p

Re: [Freesurfer] Samseg dice is low

Hi Dilek, 0.2 Dice seems extremely low. Also, in my experience, a threshold of 0.05 on T1w-FLAIR scans will show up many FP lesions. We usually reported a Dice in the range of 0.50-0.65 on various datasets (MS and WMH, default threshold, T1w-FLAIR), so I'm surprised by a such low value. Can yo

Re: [Freesurfer] samseg longitudinal processing - process killed

Hi yvg, It is most likely a memory error problem. You will need to allocate more RAM to your calls. You probably need at least 32GB or RAM for such big images (might be even more). You can process the FLAIR images with mri_robust_template since they are slighly smaller than your T1 images.

[Freesurfer] R: SAMSEG Lesion images not written

ation (seg.mgz) as a lookup table and select only the label values that you're interested in. Hope it helps, Stefano Da: Giulio Siracusano Inviato: domenica 27 novembre 2022 04:01 A: Cerri, Stefano ; freesurfer@nmr.mgh.harvard.edu Oggetto: Re: [Freesurf

[Freesurfer] R: SAMSEG Lesion images not written

using ASEG segmentation. Look at the help function for more information. Stefano Da: Giulio Siracusano Inviato: domenica 27 novembre 2022 10:07 A: Cerri, Stefano Cc: freesurfer@nmr.mgh.harvard.edu Oggetto: Re: [Freesurfer] SAMSEG Lesion images not written

[Freesurfer] [FreeSurfer] ValueError axes don't match array

External Email - Use Caution Hi Krystal, Can you send your samseg command and the output of mri_info on both the T1 and coregistered FLAIR images? It seems that the two images have different image sizes. Stefano Subject:[Freesurfer] [FreeSurfer] ValueError axes don't m

Re: [Freesurfer] Samseg cortex parcellation

Hi Valentin, Glad to hear that. Did you use the command "samseg" with the --recon flag? If yes, there are a couple of flags that you can add so that the output of SAMSEG ("seg.mgz") is used in more steps in the recon-all stream: " --fill : use samseg to create filled.mgz instead of recon-all"

Re: [Freesurfer] SAMSEG in MS terminal error

Hi Gabriel, These messages look like TensorFlow warnings. They just mean that you're running it on a system without GPUs. They should not stop the run. Is this the last message printed by run_samseg, or does the algorithm continues? Can you provide more details on when the run terminates? Stef

Re: [Freesurfer] Samseg lesion mask holes

Hi Evie, --save-probabilities and --save-posteriors will save two different things for lesions. The former is an initialization of lesions based on intensity and location constraints, and the latter is a probability map of seg.mgz after we also introduce lesion shape constraints (Cf Fig.4 of h

[Freesurfer] R: Samseg lesion mask holes

an of the WM class (default is GM), e.g.: --lesion-mask-structure White --lesion-mask-pattern 1 However, these changes might increase false positive lesions. Stefano Da: Evie Neylon Inviato: venerdì 31 marzo 2023 04:28 A: Cerri, Stefano,PHD Cc: frees

Re: [Freesurfer] Longitudinal SAMSEG with multi-contrast data

External Email - Use Caution Hi Gonzalo, Assuming that you have run mri_robust_template and obtained longitudinally registered images (e.g., tp0_t1_reg.mgz, tp1_t1_reg.mgz, etc...), then you can co-register each additional contrast of each time point with the commands: mri_core