multimodallearning / Lung250M-4B

Lung250M-4B: A Combined 3D Dataset for CT- and Point Cloud-Based Intra-Patient Lung Registration
Apache License 2.0
30 stars 3 forks source link

Generation of corrfieldFlowPcd[Tr/Ts] #4

Open norlcorxilea opened 1 month ago

norlcorxilea commented 1 month ago

Thank you very much for your publication of dataset generation! However, I still have a problem. I wonder how to generate the data in corrFieldFlowPcd[Tr/Ts] folders. In addition, in corrfield folder, I find the following line of code: nib.save(nib.Nifti1Image(img_mov_warped.cpu().numpy(), np.eye(4)), '{}.nii.gz'.format(output_path)) However, I do not find correspondent files in dataset. I wonder whether this file is related to generation of corrFieldFlowPcd[Tr/Ts], and if so, how to generate. Thanks!

eocv commented 1 month ago

I have a similar issue with the generation of the DVF. I understand that the corrField algorithm produces a NIfTI file with the displacements/DVF and a CSV file with the Förstner keypoints computed on the Fixed and Moving images. First of all, the code as it is produces a bad NIfTI file, which is not either configured as a NIfTI Transform. I tried to properly set the transform using Nibabel but even like this the resulting DVF is far from showing coherent deformations. I'm interested on the code required to properly store the DVF, with similar structure as the Fixed/Mobile images (WxHxDx3).