rwindsor1 / spinenet Goto Github PK

In the tutorial page thre is an example of NIFTI predictions whre the pixel spacing is
sx, sy, sz = image.header.get_zooms() # get pixel spacings

However when calling the detection function is specified as
vert_dicts = spnt.detect_vb(scan.volume, scan.pixel_spacing[0])

In the case tat would like to use the both pixel spacing the function detect_vb throus an error, despite in teh definition in theory allow .
def detect_vb(
self,
volume : np.ndarray,
pixel_spacing : Union[np.ndarray, List[float], torch.Tensor],
debug: bool = False,
penalise_skips: bool = True,
remove_single_slice_detections: bool = True,
) -> VertDicts:
"""
Use SpineNet to detect and label vertebral bodies in a volume.

    Parameters
    ----------
    volume : np.ndarray
        The volume to detect vertebrae in. Should have shape (height,width, number of sagittal slices).
    pixel_spacing : Union[np.ndarray, List[float], torch.Tensor]
        The pixel spacing of the volume, specifically the distance between adjacent pixels in the sagittal direction.
        This has order height, width

Numpy np.bool is deprecated since v1.20 I would suggest that you upload the file spinet/utils/gen_utils poky2
line 370 mask = np.zeros(shape, dtype=np.bool_)
I can create a merge request with the cahnge if you would like

Hi, I am currently testing SpinenetV2 on the spinegeneric dataset and I am experiencing labeling issues and missed detections for T2w images, as displayed on the following images:

Could you explain me what is wrong with those images ? And how they are different from the training dataset used for SpinenetV2 ? Because according to your Readme.md T2w contrasts should be handled.

Hi, I'm currently doing a benchmark of methods to detect vertebral discs (or vertebral bodies) on MRI scans and I truly believe that your method is interesting and could be part of this benchmark. Therefore, I would like to know if it was possible to have access to your training scripts in order to qualitatively compare this method with other deep learning based methods.