0

我正在尝试使用 GDCM 从 dicom 文件中仅读取一个切片。数据是多切片dicom。

r = gdcm.ImageRegionReader()
r.SetFileName( 'test.dcm' )
r.ReadInformation()

image = r.GetImage()
dimensions = image.GetDimensions()
pixelformat = image.GetPixelFormat()
pixelsize = pixelformat.GetPixelSize()

bytes_per_slice = dimensions[0] * dimensions[1] * pixelsize

box = gdcm.BoxRegion()
box.SetDomain(0, dimensions[0] - 1, 0, dimensions[1] - 1, 1, 1)

r.SetRegion(box)

这一切都有效，但我怎样才能读取像素？

nptypes = {gdcm.PixelFormat.UINT8  :numpy.int8,
            gdcm.PixelFormat.INT8   :numpy.uint8,
            gdcm.PixelFormat.UINT16 :numpy.uint16,
            gdcm.PixelFormat.INT16  :numpy.int16,
            gdcm.PixelFormat.UINT32 :numpy.uint32,
            gdcm.PixelFormat.INT32  :numpy.int32,
            gdcm.PixelFormat.FLOAT32:numpy.float32,
            gdcm.PixelFormat.FLOAT64:numpy.float64 }

arr = np.array(bytes_per_slice, dtype=pixelformat.GetScalarType())
# also tried np.uint8 as dtype
buffer = np.getbuffer(arr)


#
# now this fails
#
r.ReadIntoBuffer(buffer, bytes_per_slice)
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-95-a8022420e1c1> in <module>()
----> 1 r.ReadIntoBuffer(buffer, bytes_per_slice)

lib/python2.7/site-packages/gdcmswig.pyc in ReadIntoBuffer(self, inreadbuffer, buflen)
  20678         Read into buffer: false upon error
  20679         """
> 20680         return _gdcmswig.ImageRegionReader_ReadIntoBuffer(self, inreadbuffer, buflen)
  20681 
  20682 ImageRegionReader_swigregister = _gdcmswig.ImageRegionReader_swigregister

TypeError: in method 'ImageRegionReader_ReadIntoBuffer', argument 2 of type 'char *'

如何传递 numpy 缓冲区？我尝试了一些 ctype 的东西，但都没有奏效。

4

1 回答 1

0

我想到了！！

nptypes = {gdcm.PixelFormat.UINT8  :numpy.int8,
        gdcm.PixelFormat.INT8   :numpy.uint8,
        gdcm.PixelFormat.UINT16 :numpy.uint16,
        gdcm.PixelFormat.INT16  :numpy.int16,
        gdcm.PixelFormat.UINT32 :numpy.uint32,
        gdcm.PixelFormat.INT32  :numpy.int32,
        gdcm.PixelFormat.FLOAT32:numpy.float32,
        gdcm.PixelFormat.FLOAT64:numpy.float64 }

byte_buffer = np.zeros((bytes_per_slice), 
dtype=nptypes[pixelformat.GetScalarType()]).tobytes()
r.ReadIntoBuffer(byte_buffer, bytes_per_slice)

image =  np.ndarray(shape=(bytes_per_slice/2,),dtype=nptypes[pixelformat.GetScalarType()], buffer=byte_buffer)

imshow(image.reshape(dimensions[1],dimensions[0]))
于 2017-11-30T00:42:45.213 回答