Question regarding input as 3D image and setting voxelbased = False

Community Radiomics
1

Hi,

I think I just asked 2 related question but I wish I could get answers from your side both if possible.

Im dealing with 2D MRI images. For each patient, the resolution in x-y direction is higher than the z direction, which means it is a very common case – a stack of 2D images for each patient. When I input a stack of the 2D images (seen as 3D image, ) and it’s corresponding stack of 2D mask image (seen as 3D mask image) to the “.execute” function, for example:

result_vec = extractor.execute(curr_img_path, curr_mask_path, voxelBased=True/False)

Here, curr_img_path is the path to the 3D image (.nrrd format) and the curr_mask_path is the path to the 3D mask image (.nrrd format).

If I set voxelBased=True, then the returned result_vec is a set of 3D images with respect to each radiomics features, but if I set voxelBased=False, the retured result_vec is a feature vector that for each feature there is a corresponding scalar value.

My quesion:

  1. If I’d like to calculate the radiomics feature for the entire 3D image as a whole, e.g. I want the firstorder_10Percentile feature for this 3D image, do I need to set voxelBased = True? Or False?

  2. If I set Voxelbased = True, what do the returned 3D feature maps mean?

Thanks!

2

Addressed in PyRadiomics issue #600.

1 Like