Using DICOM Header Metadata in Transforms

DICaugment also supports 3D image augmentations that utilize metadata from a DICOM header file to apply both pixel-level and spatial-level transformations

This guide will walk you through the process of using DICaugment to augment 3D images using metadata from a dicom header.

1. Import the necessary modules:

   import dicaugment as dca
   

2. Read in scan and dicom header from disk

   scan, dicom = dca.read_dcm_image(
       path='path/to/dcm/folder/',
       return_header=True         # Set as True to recieve scan and dicom header
   )
   

   Alternatively, you may define the dicom header manually with a dictionary

   dicom = {
       "PixelSpacing" : (0.48, 0.48),
       "RescaleIntercept" : -1024.0,
       "RescaleSlope" : 1.0,
       "ConvolutionKernel" : 'b30f',
       "XRayTubeCurrent" : 240
   }
   

3. Define an augmentation pipeline using A.Compose:

   aug = dca.Compose(
       [
           # The Rescale Slope Intercept transformation converts the pixel values of a scan into Hounsfield Units (HU)
           # by using the `RescaleSlope` and `RescaleIntercept` values from the dicom header
           dca.RescaleSlopeIntercept(),
   
           # The Set Pixel Spacing transformation resizes each slice of the scan so that the `PixelSpacing` value
           # in the dicom header is equal to `(space_x, space_y)`
           dca.SetPixelSpacing(space_x = 0.5, space_y = 0.5),
   
           # The NPSNoise transormation applies a random change in the magnitude of the noise present in the
           # image consistent with the kernel type provided in the DICOM header.
           dca.NPSNoise(magnitude = (50,150)),
           # the magnitude of the noise will be randomly selected from the range of [50, 150]
       ]
   )
   

4. Apply the transformation to your 3D image data and mask

   As with other augmentation pipelines, transform must be called using explicit keyword arguements: image and dicom, where the scan should be passed in image and the dicom header should be passed in dicom. The output of this transformation will be a dictionary that contains the augmented scan under the key image.

   transformed_output = transform(image=scan, dicom=dicom)
   transformed_scan = transformed_output["image"]