重采样DICOM图像遇MultiValue错误：map(float)拼接列表异常求助

Hey there, let's get this sorted out—this error is super common when porting older Python 2 DICOM code to Python 3, so you’re not alone!

Why This Happens #

Two main culprits here:

1. Python 3’s map() behavior: In Python 2, map() returns a proper list, but in Python 3 it spits out an iterator. You can’t concatenate an iterator with a list directly, which triggers that error.
2. Pydicom’s MultiValue objects: Modern pydicom versions return MultiValue (a list-like subclass) for metadata like PixelSpacing. Even if you use map(), combining this with other lists can cause type mismatches in Python 3.

Step 1: Find the Problematic Line #

The error points to your map(float) call. Chances are your code has something like this:

current_spacing = map(float, scan.PixelSpacing) + [float(scan.SliceThickness)]

This worked in Python 2 because map() gave a list, but in Python 3, that map() result is an iterator—so adding it to a list throws the error.

Step 2: Fix the Conversion #

We just need to make sure we’re working with actual lists. Here are two clean ways to fix it:

Option 1: Convert map() to a list explicitly

# Turn the map result into a list first
pixel_spacing = list(map(float, scan.PixelSpacing))
# Now safely add the slice thickness
current_spacing = pixel_spacing + [float(scan.SliceThickness)]

Option 2: Use list comprehensions (more Python 3-friendly)

List comprehensions are easier to read and avoid iterator confusion entirely:

pixel_spacing = [float(val) for val in scan.PixelSpacing]
current_spacing = pixel_spacing + [float(scan.SliceThickness)]

Step 3: Update Your Full Resample Function #

Here’s how your resample function should look with the fix included (I’ve added comments for clarity):

import numpy as np
import scipy.ndimage.interpolation as interp

def resample(image, scan, new_spacing=[1,1,1]):
    # Fix: Convert DICOM spacing values to a proper list of floats
    pixel_spacing = [float(val) for val in scan.PixelSpacing]
    current_spacing = pixel_spacing + [float(scan.SliceThickness)]
    
    # Calculate resize factors (unchanged logic)
    resize_factor = np.array(current_spacing) / np.array(new_spacing)
    new_real_shape = image.shape * resize_factor
    new_shape = np.round(new_real_shape)
    real_resize_factor = new_shape / image.shape
    new_spacing = np.array(current_spacing) / real_resize_factor
    
    # Resample the image (using scipy's zoom)
    resampled_image = interp.zoom(image, real_resize_factor, mode='nearest')
    
    return resampled_image, new_spacing

Quick Check #

After making this change, test the function again. The error should disappear because we’re now concatenating two proper lists instead of an iterator/MultiValue and a list.

内容的提问来源于stack exchange，提问作者1369