Averaging PET projection data

[Rommy Elyan]

Does anyone know how to average the PET surface projection data after running the pet surface processing pipeline?

The data for each subject is in this format:

trc-18FAV45_pet_space-fsaverage_suvr-cerebellumPons_pvc-iy_hemi-lh_fwhm-5_projection.mgh

I tried getting an average through the statistics-surface pipeline but there was an np.array error.

[ninon....@gmail.com]

Hello Elyan,

You could check in the code how the regional averages are obtained (or average the regional averages if this corresponds to what you aim to obtain).

Best,

Ninon

[Rom E]

Hi Ninon,

I kinda figured it out a while ago. I ended up using nibabel and numpy in python to average the surfaces across subjects. Hopefully this function helps someone out later. 
After using this function, I used SurfStat to project the averaged files onto a fsaverage surface

import nibabel as nib

import numpy as np

def average_mgh_files(input_files, output_file):

    # Read the first file to initialize the sum and get the affine

    first_img = nib.load(input_files[0])

    sum_data = first_img.get_fdata().astype(np.float32)  # Ensure data is float32

    affine = first_img.affine

    # Iterate over the rest of the files and accumulate the sum

    for file in input_files[1:]:

        img = nib.load(file)

        sum_data += img.get_fdata().astype(np.float32)  # Ensure data is float32

    # Calculate the average

    avg_data = sum_data / len(input_files)

    # Convert the average data to a recognized data type, here using float32

    avg_data = avg_data.astype(np.float32)

    # Save the result to a new .mgh file

    avg_img = nib.MGHImage(avg_data, affine)

    nib.save(avg_img, output_file)

rh_input_files = ['ses-1_trc-18FAV45_pet_space-fsaverage_suvr-cerebellumPons_pvc-iy_hemi-rh_fwhm-5_projection.mgh', ...]
# include list of right hemisphere files for every subject, then do the same for left hemisphere

average_mgh_files(rh_input_files, 'rh_average.mgh')