By: SenticLab.UAIC on May 19, 2022, 7:27 a.m.
Hello,
I got the following error on evaluation - Failed Tolerance: 9.6100003e-07
Do you have any idea why I'm receiving this?
By: wilkiema on May 19, 2022, 8:20 a.m.
HI, Can you try to save your prediction as the below and submit your result again?
label = label.astype(np.uint8) label = sitk.GetImageFromArray(label) label.CopyInformation(image) # image loaded by "sitk.ReadImage" sitk.WriteImage(label, filename)
By: SenticLab.UAIC on May 19, 2022, 9:07 a.m.
Thank you, using the same header as the original image did the trick. I used nibabel and bellow is my code. Maybe it will help others who use the same library.
img = nib.load(3d_volume_path) seg = nib.load(seg_path).get_fdata() # segmentation numpy array seg = nib.Nifti1Image(seg, img.affine, img.header) nib.save(img, save_path)
