I cant see what you did on the "fmri model specification BUT:

1 - Have you added all the volumes ****.nii,1...****.nii,2...etc..during the preprocessing and statistics? On SPM you should to manually add all functional image volumes, always.

2 - Have you defined absolute thresholds on 1 level?

Currently, I facing the same problem.

I checked each step in the sequence and found that the problem is related to the anatomical image. when I applied the segmentation step to get the WM, GM, and CSF images it gives me black images.

Looks like SPM doesn't "find" any voxels to execute the model. This might be related to either wrong input images, or a threshold value that eliminates all voxels. I would double-check input images, check the model parameters (e.g. threshold). If the model runs with normalization this could indicate that indeed there is a threshold that eliminates all voxels (as normalization scales voxel values). As Brunno alludes to, the distinction between relative and absolute thresholds is important to consider.

Tamoor Malik my problem has been solved when i adjust manually the origin of the anatomical image as well as the functional images according to this video https://www.youtube.com/watch?v=AwNJAUKLhqY

Hi Tamoor,

I just had the same problem and I found out that the values of the parametric modulators were too high (order of 1e+10). I divided them by a common factor to make the numbers smaller and everything works fine now.

I hope you solved your problem and advanced with your project!

Maite

Doaa Mousa Maité Crespo García Lasse Bang Brunno Machado de Campos Thank You very much.... I appreciate you all

For new comers, facing the same error, i like to share my experience, actually it is because of thresholding problem due to the changed data structure, to overcome this issue i.e. " spm_est_non_sphericity ", you need to adjust Masking Thershold value during Specifying 1st Level in SPM12. for more detail... contact me anytime... Thank You

I am finding the same problem. I changed the threshold value on the specifying first level to 0 but I'm still finding the same problem! Does anyone have an idea of what is wrong?

Michelle Miranda first i will suggest you to check reorientation, secondly you may check putting normalization to scaling( which is not preferrable but sometime it works), last one is threshold which you already checked by putting 0, i think it must work, but if not so you may put minus infinity instead of 0, like this -inf

Thank You

I am conducting an ALE meta-analysis and am also facing a failed model estimation. I've tried the above recommendations as well as others, and have not had any success. I am very new to spm and am struggling to identify what the problem is. Any help would be greatly appreciated!

Kaitlyn May , According to rules, the images should be in order, so that all preprocessesing steps follow completely, otherwise you have to face errors at each end, sometime during Reorientation, when we are adjusting template with study image by MASKING... it is based on our own handling.... as shown in file (1.png and 2.png... it is SPM8 version- old assignment, referring for you to understand)... now in SPM12, when you open "CHECK REGISTER" you can open your template with study image, later right click on any image... another menu will popup... heres you can select DISPLAY>LABELS>NEURO.... from this option you are able to see accuracy more precisely... as shown in image 22 and 11... Regionwise reorientation... before i found that even after simple reorientation i got the error of "" failed model estimation... no significant voxels"" i checked my study image using label option and found the problem as shown in attached image(44.png)... myself i found it very useful and recommend this option before proceeding to preprocessing steps... you may check this with your images either they are perfectly reorientate Regionwise... GOOD LUCK

Tamoor Malik Thank you so much for your VERY helpful reply. I've been playing with reorientation today. I've been asked to use spm to perform a Monte Carlo correction on ALE data. This is a very basic question (this is my first time using spm and conducting an ALE analysis), but based on your images below it doesn't look like spm is equipped to use ALE images (see attached files for examples) since I do not have full functional images. Are these images even usable in spm? Any advice you could give me on this would be greatly appreciated.

Kaitlyn May check the attached image, of template EPI with your two study images, it seems like your study images are not appropriate , secondly according to my knowledge, normally after preprocessing steps, many other types of images are also generated such as beta images , mean image...etc, if you check MEAN image and other generated images with template image, it looks same as your study images (as shown in second attached file)... i am in doubt, Is your all images follow the same procedure... if so, it must fit to template, otherwise check the Image Acquisition procedure is same for all... THANK YOU

Kaitlyn May i checked your study images, by using OVERLAY to see where brain is located... it shows something interesting, which may helpful to you. specially the second image, which clearly shows study image may have problem... I GUESS SO

Tamoor Malik Thank you so much!! This is beginning to become more clear. I had a hunch that these images were not appropriate for spm, but given that I am new to both ALE analyses and to spm I needed help exploring this further and pinpointing the exact issue. Thank you, again, for taking the time to work through these with me.

before proceeding further, also checked type of image you are using, is it nifti4D, or other nifti format.... normally we convert nifti image into scans (.hdr/.img) format... and use the LABEL>NEUROMORPHOMETRICS to check reorientation, as shown in attached image... GOOD LUCK