using MATLAB to find area of flourescent region

I have a fluorescent organelle image and i need to find the area of one part of the organelle. How can i get that area?, i tried thresholding but the illlumination is super uneven. I know cropping is my best bet but I need it to be automated for multiple images where the organelle may not be in the same portion of the crop. I was pretty close when I used the following after grayscaling.

enh = imadjust(smooth, stretchlim(smooth, [0.01 0.99]), []);
p = 0.73;
T = quantile(enh(:), p);
BW = enh > T;
imshow(BW);
imfill(BW, "holes");
BW = bwareafilt(BW, 1);

I still cannot fill the cavity completely, but the specific organelle area is the most flourescent its just the holes I am having an issue with. Or is there a way to completely scrap this method and improve my thresholding for the image. I also tried imclose as well but still I had issues.

[Walter Roberson]

Please attach one of your images. Please attach a second image showing the outline of the organelle you want to select.

[Chiedza]

I have highlighted the region in yellow

[Walter Roberson]

At most I see a vague hint that maybe there might be something in the area, but that vague hint is on the order of the dirt on my display, so I don't know if there is anything there or not. The only thing I see in that image is a green-ish dot about 1/4 of the way down and about 2./3 of the way across.

[Chiedza]

Heres a better visual of the image enhanced, trying to get the round area only not the trunk

Answer 1

[Mathieu NOE]

hello

this is a quick and dirty (potential) solution :

• that does not need the Image P Tbx
• generated by a non expert in the field...

but I'm always keen to try things

so let's go - assuming your have already converted the original RGB image into either a grayscale (or just take the green channel as it's anyway a green picture)

% load image
A = double(rgb2gray(imread('image2.png')));

% centroid of the "high amplitude area"
[m,n] = size(A);
[y,x] = find(A>max(A(:))-16);
xc = mean(x);
yc = mean(y);

% image gradient
iG = abs(gradient(A));
ind = find(iG>0.25*max(iG(:))); % the threshold value is critical to the result
[yb,xb] = ind2sub(size(iG),ind);

% some coordinates conversions....
xbs = xb - xc;
ybs = yb- yc;
[th,r] = cart2pol(xbs,ybs);

% a trick to find the inner boundary (by taking
% the inverse of the radius)
[xtmp,ytmp] = pol2cart(th,1./r);
 k = boundary(xtmp,ytmp,0);
[th,r] = cart2pol(xbs(k),ybs(k));
% take unique  values (and sort theta)
[th,ia,ib] = unique(th);
r = r(ia);
% interpolation in polar coordinates
N = 50;
thi = (-pi+2*pi/N:2*pi/N:pi);
ri = interp1(th,r,thi,'spline');
[xbs,ybs] = pol2cart(thi,ri);
xb = xbs + xc;
yb = ybs + yc;
% close the boundary
xb(end+1) = xb(1);
yb(end+1) = yb(1);

figure
imagesc(A)
hold on 
plot(xc,yc,'+m','markersize',15)
plot(xb,yb,'*-r','markersize',5)

●

% use inpolygon to extract the area of interest
[X,Y] = meshgrid((1:n),(1:m));
in = inpolygon(X,Y,xb,yb);
figure
imagesc(A.*in)
% map = [0 0 0;parula(256)]; % display NaN values (points outside the polygon) in black
map = [1 1 1;parula(256)]; % display NaN values (points outside the polygon) in white
colormap(map)