Creating a 5*5 window around the initial seed point in a region growing algorithm?

Hi,

I am trying to implement a region growing algorithm to segment tumor in gastric. I would select the initial seed point manually. Then a 5*5 window would be created around the initial seed point. The minimum value inside the window would be selected as the first seed point of the region growing algorithm. Another new 5*5 window would be created around the the first seed point and the algorithm would be applied within that window. My problem is that I am unable to create the window around the initial seed point. I tried the code below but I don't get the correct result. The original image is attached as Original_Image.png and the tumor region would supposedly look as shown in Tumor.png. Any suggestions would be helpful. Thank you.

i=rgb2gray(imread('Original_Image.png'));
figure, imshow(i,[])
% gg = uint8(255 * mat2gray(i));
[x,y]=getpts;x=round(x);y=round(y);
a=imgaussfilt(i,2);
% a=rgb2gray(a);
b=adapthisteq(a);
m=regMLT(b,x,y,12);
m=imfill(m,'holes');

bw=imbinarize(m);
bw=bwareafilt(bw,1);
seg = region_seg(m, bw, 30);
seg=double(Y).*double(seg);
figure, imshow(seg,[])

Answer 1

[Maneet Kaur Bagga]

Hi Warid,

As per my understanding, the code provided is using a gaussian filter which can result in extending the window beyond the boundaries of the image, leading to errors.

Please refer to the modified code below, it will create a circular window around the initial seed point with a radius of 2 pixels. Also it checks if the window lies within the image boundaries, it uses new minimum value as the seed point for the region growing algorithm for gaining more accuracy.

% Initialize the minimum value and its corresponding coordinates
min_value = inf;
min_x = 0;
min_y = 0;

% Create a circular window around the initial seed point
radius = 2;
for i = x-radius:x+radius
    for j = y-radius:y+radius
        if i >= 1 && i <= size(i, 1) && j >= 1 && j <= size(i, 2)
            % Calculate the distance from the pixel to the seed point
            distance = sqrt((i - x)^2 + (j - y)^2);

            % Update the minimum value and its coordinates if necessary
            if distance <= radius && i ~= x || j ~= y
                if i < min_value
                    min_value = i;
                    min_x = i;
                    min_y = j;
                end
            end
        end
    end
end

% Use the new minimum value as the seed point for the region growing algorithm
a=imgaussfilt(i,2);
b=adapthisteq(a);
m=regMLT(b,min_x,min_y,12);
m=imfill(m,'holes');

Hope this helps!