Algorithm tryout error: `The image produced in 'images/inferior-alveolar-canal' is not valid`

Hi! I'm trying out the algorithm, and I have encountered the following problem: The image produced in 'images/inferior-alveolar-canal' is not valid.

It doesn't say what is causing this problem. Might be the name? Anyone have the same issue? (the stdout and stderr have not clarified anything for me. Everything looks ok). Thanks.

Best, Vicent

Hi Vicent, just some quick questions to better understand why this happens: - is the submitted code based on the example code we have provided on our GitHub repository? - which is the dtypeof the tensor you are saving?

Regards, Luca

Hi Luca,

No, the code I implemented wasn't 100% based on the one in the github repo. Precisely, my predictfunction didn't return nothing, it just saved the result. This is because my approach required the file conversion from .mha to .nii.gz back to .mha (with a slicing of the 3D images into 2D images along the three axes-- axial, saggital and coronal-- in the middle). Now I have changed the code so that predict returns a sitk.Image object as the code in your repo does. I hope this solves it.

Maybe, to anticipate any further errors, you could tell me where the error occurs? I cannot find any raiseError instance such as the one I'm getting in the evalutils code nor in the ToothFairy github repo.

Thanks. Vicent

Hey,do you solve the question?I test my algorithm on the page and get the same error.

actually the dtype of the output should be "uint8".....

Dear zt081508,

Yes, I solved my error by forcing the algorithm to output a segmentation as an array in dtype np.uint8 as you mention in your comment.

Best regards, Vicent