dicom_decompressor.py

import pydicom as PDCM
import numpy as np
import cv2 as cv
import glob
import os
import shutil 


def conversion(Path):
    dicom_image = PDCM.read_file(Path)

#getting rows, columns, instance and window size info from dicom file tag 
#the number inside the parenthesis is associated with the tag name varibale

    rows = dicom_image.get(0x00280010).value #obtain number of rows from the tag
    columns = dicom_image.get(0x00280011).value #obtain number of colums from the tag
    dicom_instance = int(dicom_image.get(0x00200013).value) #get slice instance from the tag
    Window_Center = int(dicom_image.get(0x00281050).value) #window dimentions
    Window_Width = int(dicom_image.get(0x00281051).value) 
    window_high = int(Window_Center + Window_Width / 2)
    window_low = int(Window_Center - Window_Width / 2)


#rescale intercept and slope for hu

    if (dicom_image.get(0x00281052) is None):
        Rescale_Intercept = 0
    else:
        Rescale_Intercept = int(dicom_image.get(0x00281052).value)
    if (dicom_image.get(0x00281053) is None):
        Rescale_Slope = 1
    else:
        Rescale_Slope = int(dicom_image.get(0x00281053).value)

    updated_image = np.zeros((rows, columns), np.uint8) #duplicating the rows and columns with zeros 
    pxl_arr = dicom_image.pixel_array
    

    for each_row in range(0, rows):
        for each_col in range(0, columns):
            
            val = pxl_arr[each_row][each_col]
            rescaled_val = val * Rescale_Slope + Rescale_Intercept

            if (rescaled_val > window_high): #if the intensity is > than the window max size
                updated_image[each_row][each_col] = 255
            elif (rescaled_val < window_low): #if the intensity is < than the window min size 
                updated_image[each_row][each_col] = 0
            else:
                updated_image[each_row][each_col] = int(((rescaled_val - window_low) / (window_high - window_low)) * 255) #Normalize the intensities

                
    return updated_image, dicom_instance

def load_all_dcm(): #selecting dcm extension type only
    files = glob.glob("./covidimages/**/*.DCM", recursive = True) #getting all the dcm images 
    return files

def ignore_dcm(): #ignore the dcm extension type, so we are left with png only
    shutil.copytree("covidimages", "finaloutput", ignore = shutil.ignore_patterns('*dcm'))


if __name__ == "__main__": #first we run the conversion to get png img next to each xray dcm image, then we ignore the dcm file extension so we are left with just png images
    

    img_path = load_all_dcm()

    counter = 0    
    
    for each_row in range(0, len(img_path)):

        Output_Image, dicom_instance = conversion(img_path[each_row])
        print(img_path[each_row])
        path_slice = img_path[each_row][:-4:] #-4 or more, use it when actually woring with the covid images
        cv.imwrite((path_slice) + '_image' + str(counter) + '.png', Output_Image)
        counter = counter + 1
     
    print("done with the conversion")
    print("starting")
    ignore_dcm()
    print("Done")