first commit

DAB_Analysis_Functions.py

@@ -0,0 +1,101 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""
+Created on Fri Dec  8 15:21:34 2023
+
+Class related to analysis of DAB stained images
+
+@author: jbeckwith
+"""
+import numpy as np
+import cv2
+
+class DAB():
+    def __init__(self):
+        self = self
+        return
+
+    def imread(self, file):
+        """ imread function
+        takes RGB image and corrects openCV's ordering
+        ================INPUTS============= 
+        file is file path
+        ================OUTPUT============= 
+        img is opencv image """
+        image = cv2.imread(file) # read in image
+        img = cv2.cvtColor(image, cv2.COLOR_BGR2RGB) # make it the ordering everyone else uses
+        return img
+
+    def im2double(self, img):
+        """ im2double function
+        takes image and normalises to double
+        ================INPUTS============= 
+        img is image object
+        ================OUTPUT============= 
+        imdouble is numpy array """
+        info = np.iinfo(img.dtype) # Get the data type of the input image
+        imdouble = img.astype(np.float32) / info.max # Divide all values by the largest possible value in the datatype
+        return imdouble
+
+    def getMeanchannelvalues(self, image, mask):
+        """ getMeanLABvalues function
+        takes image, image mask and gets mean channel components 
+        assumes Lab colour space
+        ================INPUTS============= 
+        image is numpy array
+        mask is numpy logical array
+        ================OUTPUT============= 
+        imdouble is numpy array """
+        LMean = np.nanmean(np.multiply(image[:, :, 0], mask)) # mean of only the pixels within the masked area.
+        aMean = np.nanmean(np.multiply(image[:, :, 1], mask)) # mean of only the pixels within the masked area.
+        bMean = np.nanmean(np.multiply(image[:, :, 2], mask)) # mean of only the pixels within the masked area.
+        return LMean, aMean, bMean
+
+    def colourFilterLab(self, image, initial_params, rate=[0.75, 4], percentage=0.075, maxiter=30):
+        """ colourFilterLab function
+        takes image, and uses initial parameters and rate to separate out
+        coloured objects
+        ================INPUTS============= 
+        image is numpy array
+        initial_params are initial Lmean, aMean, bMean and threshold parameters
+        rate is how fast to change the parameters per iteration
+        percentage is the percentage of pixels we worry about in the mask
+        maxiter is how many iterations of optimisation to do (default 30)
+        ================OUTPUT============= 
+        image_mask is logical array of image mask
+        current_params are image analysis final params """
+        mask_current = np.zeros(image[:,:,0].shape);
+        LMean = initial_params[0];
+        aMean = initial_params[1];
+        bMean = initial_params[2];
+        thres = initial_params[3];
+        ratep_1 = rate[0];
+        ratep_2 = rate[1]; 
+
+        iteration = 1
+        while (np.nansum(mask_current) < np.multiply(percentage, image[:,:,0].size)) and (iteration <= maxiter):
+            deltaL = np.subtract(image[:,:,0], LMean) # get mean channel 1
+            deltaa = np.subtract(image[:,:,1], aMean) # get mean channel 2
+            deltab = np.subtract(image[:,:,2], bMean) # get mean channel 3
+            deltaE = np.sqrt((deltaL**2 + deltaa**2 + deltab**2)) # get change versus iteration
+            mask_previous = mask_current # store previous mask
+            mask_current = 1*(deltaE <= thres) # update mask
+            mask_current = np.where(mask_current==0, np.nan, mask_current)
+            LMean, aMean, bMean = self.getMeanchannelvalues(image, mask_current) # get new means
+            LMean = np.multiply(LMean, ratep_1) # adjust means with rate of change
+            aMean = np.multiply(aMean, ratep_1) # adjust means with rate of change
+            bMean = np.multiply(bMean, ratep_1) # adjust means with rate of change
+
+            dEmask = np.multiply(deltaE, mask_current)
+            meanMaskedDeltaE  = np.nanmean(dEmask)
+            stDevMaskedDeltaE = np.nanstd(dEmask)
+            thres = np.add(meanMaskedDeltaE, np.multiply(ratep_2, stDevMaskedDeltaE))
+            iteration += 1
+
+        image_mask = mask_previous # update mask
+        image_mask[np.isnan(image_mask)] = 0
+        current_params = np.array([LMean, aMean, bMean, thres])
+        return image_mask, current_params
+
+    def analyse_DAB_and_cells(self, file):
+        return