SimpleITK Filtering with sitk.ConnectedThresholdImageFilter() wrong output

I have tried many things to segment gray_matter, white_matter and cs_fluid images using sitk.ConnectedThresholdImageFilter(). unfortunately, I couldn't. Please, let me know what I am doing wrong.

Here is the example code:

data_dir = "<path to data dir>"
image_dir = data_dir + "images/" 
image_filenames = sorted(glob.glob(image_dir + '*.nii.gz'))

mask_dir = data_dir + "masks/" 
mask_filenames = sorted(glob.glob(mask_dir + '*.nii.gz'))

image_filename = image_filenames[0]
mask_filename = mask_filenames[0]

image = sitk.ReadImage(image_filename)
mask = sitk.ReadImage(mask_filename)

print("image_filename:", image_filename)
print("Image:")
display_image(image)
print("Mask:")
display_image(mask)

masked_image = sitk.Mask(image, mask)
print("Masked image:")
display_image(masked_image)

gm_filter = sitk.ConnectedThresholdImageFilter()
gm_filter.SetLower(1)  # Lower threshold for GM intensities
gm_filter.SetUpper(100)  # Upper threshold for GM intensities
gm_image = gm_filter.Execute(masked_image)
print("GM image:")
display_image(gm_image)

And, here is the output:

Image:

Mask:

Masked image:

GM image:

It's all good up to "GM image". I really couldn't figure out what I am doing wrong with the sitk.ConnectedThresholdImageFilter() method. Thanks in advance.

You haven't specified the seed point for the region growing, provide a seed point within the region of interest.

You can do that with the AddSeed method of the ConnectedThresholdImageFilter

edit: lets try the AddSeed method

For example:

from glob import glob
import SimpleITK as sitk

data_dir = "<path to data dir>"
image_dir = data_dir + "images/" 
image_filenames = sorted(glob(image_dir + '*.nii.gz'))

mask_dir = data_dir + "masks/" 
mask_filenames = sorted(glob(mask_dir + '*.nii.gz'))

image_filename = image_filenames[0]
mask_filename = mask_filenames[0]

image = sitk.ReadImage(image_filename)
mask = sitk.ReadImage(mask_filename)

masked_image = sitk.Mask(image, mask)

gm_filter = sitk.ConnectedThresholdImageFilter()
gm_filter.SetLower(1)
gm_filter.SetUpper(100)
gm_filter.AddSeed((50, 50, 50))
gm_image = gm_filter.Execute(masked_image)

I could do something similar to what I needed by using methods called sitk.BinaryThreshold() and sitk.And() together.

lower_threshold, upper_threshold = 100, 200 # Thresholds for gray matter
gm_image = sitk.BinaryThreshold(image, lower_threshold, upper_threshold, 1, 0)
gm_image = sitk.And(gm_image, mask)

print("GM image:")
display_image(gm_image)

Here's the result for the GM image:

I still wonder what the problem with sitk.ConnectedThresholdImageFilter() is.