can anyone please help me solve the accuracy verification issue with region props? is the performance optimal for measurements? has it been converted to square microns?

Question

Laura 2023-4-22

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1951428-can-anyone-please-help-me-solve-the-accuracy-verification-issue-with-region-props-is-the-performanc

编辑： Laura 2023-4-25

8_controlmatlab.png

in my current code the user makes a perfect circle over the completely black region in the image. From the edges of the circle 12 equally distributed "bins" or regions are segmented to the edges of the image. however i would like to verify that the emasuremnts being performed are accurate.

for m = 1:numfiles
    fprintf('Processing file #%d of %d : "%s".\n', m, numfiles, allFileNames{m});
    MyRGBImage = fullfile(pwd, allFileNames{m});
    % Read in image.
    imageData = imread(MyRGBImage);
    
    MI = imread(MyRGBImage,1); %%AI channel mitochondria
    
    AI = imread(MyRGBImage,2); %%BI channel LD
    LD = imread(MyRGBImage,3);  %% just actin
    % % %     actin, mitochondria and lipids
    grayImg = MI + LD + AI;
    [rows, columns, numberOfColorChannels] = size(grayImg);
    %% Mask for dark regions removal
    zgrayImg = im2double(grayImg);
    
    II = zscore(zgrayImg);
    
    % remove speckle
    D = wiener2(II,[55,55],10000000/6);  
    DD = imbinarize(D,-0.2);
    
    %% Uncomment when measuring mitochondria
    %% Change between channels by replacing MI, LD or AI as desired, Apply background subtraction, initialized at 25    
    bg = imopen(MI, strel('disk', 25));
    
    imgNoBg = MI - bg; % change this to the channel to be measured
    imgContrast = imadjust(imgNoBg);
    % For mitochondria channel 
    edges = edge(imgContrast,'Canny'); 
    se = strel('disk', 1);
    edgesClean = imclose(edges, se);
    edgesClean = imfill(edgesClean, 'holes');
    for k = 1:numZones
        % Create a binary mask for this zone
        %     zoneMask = (sqrt((X - center(1)).^2 + (Y - center(2)).^2) >= radius(k)) & (sqrt((X - center(1)).^2 + (Y - center(2)).^2) < radius(k+1));
        %     areaOfEachZone(k) = sum(zoneMask, 'all') * pixelSize^2;
        if k == 1
            % Create a mask for the first zone based on distance from center
            zoneMask = sqrt((X - centerc(1)).^2 + (Y - centerc(2)).^2) < radii(1);
        else
            % Create a mask for this zone based on distance from center
            zoneMask = (sqrt((X - centerc(1)).^2 + (Y - centerc(2)).^2) > radii(k-1)) & (sqrt((X - centerc(1)).^2 + (Y - centerc(2)).^2) < radii(k));
        end
        % Apply functional mask to the binary mask for this zone
        %     zoneMask = zoneMask .* im2double(edgesClean) .* DD;
        areaOfEachZone(k) = sum(zoneMask, 'all') * pixelSize^2;
        zoneMask = zoneMask .* im2double(edgesClean) .* DD;
        % Calculate connected components in this zone
        cc = bwconncomp(zoneMask);
        
        % Calculate region properties for each connected component in this zone
        statscc = regionprops(cc, 'Area', 'Centroid', 'Eccentricity', 'Perimeter', 'MajorAxisLength', 'MinorAxisLength','Circularity');
        % Calculate features for this zone
        numObjects = cc.NumObjects;
        if numObjects > 0
            areas = [statscc.Area];
            profileCounts(k) = numObjects;
            totalArea(k) = sum(areas) * pixelSize^2;
            zoneArea(k) = areaOfEachZone(k);
            % Calculate average size of mitochondria for this zone
            avgSize(k) = mean(areas) * pixelSize^2;
            circularities = [statscc.Circularity];
            circularities(circularities > 1) = 1; % Cap circularities greater than 1 at 1
            avgCircularity(k) = mean(circularities(isfinite(circularities)));
            ferets = [statscc.MajorAxisLength];
            avgFeret(k) = mean(ferets) * pixelSize;
            minFerets = [statscc.MinorAxisLength];
            avgMinFeret(k) = mean(minFerets) * pixelSize;
        end
        
    end

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Image Analyst 2023-4-22

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1951428-can-anyone-please-help-me-solve-the-accuracy-verification-issue-with-region-props-is-the-performanc#answer_1221023

You didn't post the complete file. Can you attach it? What is numZones? 12? Perhaps the other zones are off the end of the image.

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Laura 2023-4-25

在 MATLAB Online 中打开

@Image Analyst Attached and below. For the last three months or so I have been trying to verify the measurements of my code are accurate. i set the pixelsize to 1 and then later i convert the measurements to square microns. How can i very that my codes measurements are accurate of the images being measured?

for m = 1:numfiles
    fprintf('Processing file #%d of %d : "%s".\n', m, numfiles, allFileNames{m});
    MyRGBImage = fullfile(pwd, allFileNames{m});
    % Read in image.
    imageData = imread(MyRGBImage);
    
    MI = imread(MyRGBImage,1); %%AI channel mitochondria
    
    AI = imread(MyRGBImage,2); %%BI channel LD
    LD = imread(MyRGBImage,3);  %% just actin
    % % %     actin, mitochondria and lipids
    grayImg = MI + LD + AI;
    [rows, columns, numberOfColorChannels] = size(grayImg);
    %% Mask for acinus removal
    zgrayImg = im2double(grayImg);
    
    II = zscore(zgrayImg);
    
    % remove speckle
    D = wiener2(II,[55,55],10000000/6);  
    DD = imbinarize(D,-0.2);
    
    %% Uncomment when measuring mitochondria
    %% Change between channels by replacing MI, LD or AI as desired, Apply background subtraction, initialized at 25    
    bg = imopen(MI, strel('disk', 25));
    
    imgNoBg = MI - bg; % change this to the channel to be measured
    imgContrast = imadjust(imgNoBg);
    % For mitochondria channel 
    edges = edge(imgContrast,'Canny'); 
    se = strel('disk', 1);
    edgesClean = imclose(edges, se);
    edgesClean = imfill(edgesClean, 'holes');
    for k = 1:numZones
        % Create a binary mask for this zone
        %     zoneMask = (sqrt((X - center(1)).^2 + (Y - center(2)).^2) >= radius(k)) & (sqrt((X - center(1)).^2 + (Y - center(2)).^2) < radius(k+1));
        %     areaOfEachZone(k) = sum(zoneMask, 'all') * pixelSize^2;
        if k == 1
            % Create a mask for the first zone based on distance from center
            zoneMask = sqrt((X - centerc(1)).^2 + (Y - centerc(2)).^2) < radii(1);
        else
            % Create a mask for this zone based on distance from center
            zoneMask = (sqrt((X - centerc(1)).^2 + (Y - centerc(2)).^2) > radii(k-1)) & (sqrt((X - centerc(1)).^2 + (Y - centerc(2)).^2) < radii(k));
        end
        % Apply functional mask to the binary mask for this zone
        %     zoneMask = zoneMask .* im2double(edgesClean) .* DD;
        areaOfEachZone(k) = sum(zoneMask, 'all') * pixelSize^2;
        zoneMask = zoneMask .* im2double(edgesClean) .* DD;
        % Calculate connected components in this zone
        cc = bwconncomp(zoneMask);
        
        % Calculate region properties for each connected component in this zone
        statscc = regionprops(cc, 'Area', 'Centroid', 'Eccentricity', 'Perimeter', 'MajorAxisLength', 'MinorAxisLength','Circularity');
        % Calculate features for this zone
        numObjects = cc.NumObjects;
        if numObjects > 0
            areas = [statscc.Area];
            profileCounts(k) = numObjects;
            totalArea(k) = sum(areas) * pixelSize^2;
            zoneArea(k) = areaOfEachZone(k);
            % Calculate average size of mitochondria for this zone
            avgSize(k) = mean(areas) * pixelSize^2;
            circularities = [statscc.Circularity];
            circularities(circularities > 1) = 1; % Cap circularities greater than 1 at 1
            avgCircularity(k) = mean(circularities(isfinite(circularities)));
            ferets = [statscc.MajorAxisLength];
            avgFeret(k) = mean(ferets) * pixelSize;
            minFerets = [statscc.MinorAxisLength];
            avgMinFeret(k) = mean(minFerets) * pixelSize;
        end
        
    end