This function reads in cell images and parameters (lower bleb size threshold, upper bleb size threshold and the alpha threshold) and outputs "blebdata" which is a structure array containing the following fields:
- blebdata.bleb(:,:,j,i) = binary matrix for bleb j at time frame i
- blebdata.blebsize(:,i) = array of blebsizes
- blebdata.alpha(:,i) = array for the "roundness" of cell
- blebdata.showAllBlebs(:,:,i) = single binary matrix for all the blebs at time frame i
- blebdata.numberofblebs(i) = number of blebs at time frame i
- blebdata.numberofimages = total number of images
- blebdata.indivBlebEdge(:,:,j,i) = matrix of 1s 2s and 3s representing the blebbing intensities of bleb j at time frame i
- blebdata.outline(:,:,i) = binary matrix representing the outline of the cell
- blebdata.images.intensities(:,:,i) = grayscale image matrix for time frame i
- blebdata.images.BW(:,:,i) = black and white image matrix for time frame i
- blebdata.images.BW2(:,:,i) = black and white image matrix after filling in the cell and removing isolated pixels
Using findBlebs.m, this script finds all connected components of size at least 50 pixels. We then can go through and manually delete connected components that are not blebs. In order to do this, use arrow keys to loop through the images and use the number keys to delete the corresponding detected blebs. To save the variables into a .mat press the zero key.
This script calculates the training and validation accuracy for our definition of a bleb. Also included, is the computation of a Gaussian Bayes classifier. As noted in our report and this webpage, our simple model's accuracy was better than that of the Bayes classifier.