我正在尝试使用 image_slicer 函数将 DICOM 图像分割成图 block ,但它无法识别 DICOM。
我已经阅读了 DICOM 并将它们转换为 np 数组:
dcm_文件[0]
array([[-1024, -1024, -1024, ..., -1024, -1024, -1024],
[-1024, -1024, -1024, ..., -1024, -1024, -1024],
[-1024, -1024, -1024, ..., -1024, -1024, -1024],
...,
[-1024, -1024, -1024, ..., -1024, -1024, -1024],
[-1024, -1024, -1024, ..., -1024, -1024, -1024],
[-1024, -1024, -1024, ..., -1024, -1024, -1024]], dtype=int16)
并且能够通过以下方式查看图像:
from PIL import Image
import numpy as np
img = Image.fromarray(dcm_files[0])
img.show()
然后尝试切片:
import image_slicer
image_slicer.slice(img, 64)
错误:“图像”对象没有属性“读取”
谢谢!
请您参考如下方法:
image_slicer 模块使用文件名,而不是文件实例。因此,您必须将数组保存到某种本地化。您可以使用 tempfile module为了这个目的。
from PIL import Image
import numpy as np
import image_slicer
import tempfile
array = np.random.randint(0,200, size=(128,128), dtype='int32')
img = Image.fromarray(array)
temp = tempfile.NamedTemporaryFile()
img.save(temp.name, format="png")
image_slicer.slice(temp.name, 4)
#if you want to play with the slices:
tiles = image_slicer.slice(temp.name, 4, save=False)
