How to automatically identify abnormal regions and highlight boundaries in a different color

I have a set of images for classification.

For this I have already extracted features from which the images have been already segmented and divided into four quadrants.

Now I want to use this feature values to idenifty the quadrants with the specified value ranges given below.

Here is my code;

% Read in the images. 
I= imread ('nhimg-*.jpg'; 
 % Segmentation of the ROI. 
[S] = LungMask (I);    
% Dividing Segmented ROI into quadrants. 
[RUQ, LUQ, RLQ, LLQ] = crop(S); 
% Extracting Features from each quadrant of the segmented images.  
[Stats1, MaxProb1, Entropy1] = Prop(RUQ)
[Stats2, MaxProb2, Entropy2] = Prop(LUQ)
[Stats3, MaxProb3, Entropy3] = Prop(RLQ)
[Stats4, MaxProb4, Entropy4] = Prop(LLQ)  
.
.
.
.

Now for this program I need to add the below feature ranges for each qaudrant to check whether the qaudrant features are in the defined range and if so mark the border of the respective quadrant in red color and create a label as "Abnormal".

These are the feature ranges;

Feature RUQ LUQ RLQ LLQ

Contrast 18-150 50-70 7-157 21-179

Correlation 0.97-0.99 0.97-0.99 0.98-1.00 0.97-0.99

Energy 0.30-0.54 0.32-0.48 - -

Etropy 2.79-3.76 2.85-3.94 - -

Max.Prob 0.58- 0.84 - - -

When the images are processed with this I need a final output like this image attached.

Can you help me to automize this existing code to get the output I need? Thanks in advance.

[darova]

What is the criteria of abnormal image?

[Nileema Abedeera]

The above given five feature value ranges are the criteria.

Contrast, Correlation, Energy, Entropy and Max.Prob. The given values ranges can only be seen in abnormal images.

[darova]

if ~isempty(MaxProb)
    image = abnormal
end

[Nileema Abedeera]

Hi Darova, thank you.

But I didn't get you!

Would you explain this please?

Thanks again.

[darova]

Can you attach the data for my experiments?

Answer 1

[Image Analyst]

You have to help us first. You forgot to attach 'nhimg-*.jpg' plus you forgot to attach the code for your LungMask(), crop(), and Prop() functions. What can we do without all those? Anyway, did you try subplot()M title(), and xlabel()?

subplot(2, 2, 1);
imshow(RUQ);
title('RUQ', 'FontSize', 15);
if IsAbnormal(RUQ)  % You need to write this function
    xlabel('Abnormal', 'FontSize', 15)
end
subplot(2, 2, 2);
imshow(LUQ);
title('LUQ', 'FontSize', 15);
if IsAbnormal(LUQ)  % You need to write this function
    xlabel('Abnormal', 'FontSize', 15)
end
subplot(2, 2, 3);
imshow(RLQ);
title('RLQ', 'FontSize', 15);
if IsAbnormal(RLQ)  % You need to write this function
    xlabel('Abnormal', 'FontSize', 15)
end
subplot(2, 2, 4);
imshow(LLQ);
title('LLQ', 'FontSize', 15);
if IsAbnormal(LLQ)  % You need to write this function
    xlabel('Abnormal', 'FontSize', 15)
end

[Nileema Abedeera]

Thank you Image Analyst.

And yes I tried few methods but didn't work.

Here I have attached the .jpg image.

Kindly help me get the output required as mentioned in the question above. Thank you so much.

[Nileema Abedeera]

Hi again,

I have tried for getting this output with my code below.

Kindly help me to understand whether this code dveloped by myslef is correct for what I want to get as the output.

I = imread('NHIMG-*.jpg');
S = LungMask(I);
figure, [RUQ, LUQ, RLQ, LLQ] = crop(S);
[Stats1, MaxProb1, Entropy1] = Prop(RUQ);
[Stats2, MaxProb2, Entropy2] = Prop(LUQ);
[Stats3, MaxProb3, Entropy3] = Prop(RLQ);
[Stats4, MaxProb4, Entropy4] = Prop(LLQ);
T1 = [Stats1.Contrast, Stats1.Correlation, Stats1.Energy, Stats1.Homogeneity];
T2 = [Stats2.Contrast, Stats2.Correlation, Stats2.Energy, Stats2.Homogeneity];
T3 = [Stats3.Contrast, Stats3.Correlation, Stats3.Energy, Stats3.Homogeneity];
T4 = [Stats4.Contrast, Stats4.Correlation, Stats4.Energy, Stats4.Homogeneity];
RUQCN = Stats1.Contrast;
RUQCR = Stats1.Correlation;
RUQEG = Stats1.Energy;
RUQMP = MaxProb1;
RUQEN = Entropy1;
LUQCN = Stats2.Contrast;
LUQCR = Stats2.Correlation;
LUQEG = Stats2.Energy;
LUQEN = Entropy2;
RLQCN = Stats3.Contrast;
RLQCR = Stats3.Correlation;
RLQEN = Entropy3;
LLQCN = Stats4.Contrast;
LLQCR = Stats4.Correlation;
for Abnormal_RUQ = (RUQCN >=18 & RUQCN <= 146)| (RUQCR >= 0.97 & RUQCR <= 0.99)| (RUQEG >= 0.31 & RUQEG <=0.51)|...
(RUQEN >= 2.91 & RUQEN <=3.79) | (RUQMP >= 0.57 & RUQMP <= 0.69)
figure, subplot(2,2,1)
imshow(RUQ), title('RUQ')
if Abnormal_RUQ
xlabel('Abnormal', 'FontSize', 15)
end
end
for Abnormal_LUQ = (LUQCN >= 50 & LUQCN <= 78) | (LUQCR >= 0.97 & LUQCR <= 0.99) | (LUQEG >= 0.34 & LUQEG <= 0.52)...
(LUQEN >= 2.91 & LUQEN <= 3.79)
subplot(2,2,2), imshow(LUQ), title('LUQ')
if Abnormal_LUQ 
xlabel('Abnormal', 'FontSize', 15)
end
end
for Abnormal_RLQ = (RLQCN >= 17 & RLQCN <=157) | (RLQCR >= 0.98 & RLQCR <= 1.00) 
subplot(2,2,3), imshow(RLQ), title('RLQ')
if Abnormal_RLQ 
xlabel('Abnormal', 'FontSize', 15)
end
end
for Abnormal_LLQ = (LLQCN >= 21 & LLQCN <= 179) | (LLQCR >= 0.97 & LLQCR <= 0.99)
subplot(2,2,4), imshow(LLQ), title('LLQ')
if Abnormal_LLQ 
end
end 

What I want as the output is to label the only quadrant/s which I have specified with the given parameters, and also I want them to be highlighted witha red border.

Thank you so much for your help.

[Image Analyst]

You don't / shouldn't make a boolean and then use it after the "for" and then check it again inside the for loop.

Try it like this:

% Initialization steps.
clc;    % Clear the command window.
fprintf('Beginning to run %s.m ...\n', mfilename);
close all;  % Close all figures (except those of imtool.)
clear;  % Erase all existing variables. Or clearvars if you want.
workspace;  % Make sure the workspace panel is showing.
format short g;
format compact;
fontSize = 15;

% Have user browse for a file, from a specified "starting folder."
% For convenience in browsing, set a starting folder from which to browse.
startingFolder = pwd;  % or 'C:\wherever';
if ~exist(startingFolder, 'dir')
	% If that folder doesn't exist, just start in the current folder.
	startingFolder = pwd;
end
% Get the name of the file that the user wants to use.
defaultFileName = fullfile(startingFolder, 'NHIMG*.*');
[baseFileName, folder] = uigetfile(defaultFileName, 'Select a file');
if baseFileName == 0
	% User clicked the Cancel button.
	return;
end
fullFileName = fullfile(folder, baseFileName)

I = imread(fullFileName);
S = LungMask(I);
h2 = figure
[RUQ, LUQ, RLQ, LLQ] = crop(S);
h2.WindowState = 'maximized';
drawnow;
[Stats1, MaxProb1, Entropy1] = Prop(RUQ);
drawnow;
[Stats2, MaxProb2, Entropy2] = Prop(LUQ);
drawnow;
[Stats3, MaxProb3, Entropy3] = Prop(RLQ);
drawnow;
[Stats4, MaxProb4, Entropy4] = Prop(LLQ);
drawnow;
T1 = [Stats1.Contrast, Stats1.Correlation, Stats1.Energy, Stats1.Homogeneity];
T2 = [Stats2.Contrast, Stats2.Correlation, Stats2.Energy, Stats2.Homogeneity];
T3 = [Stats3.Contrast, Stats3.Correlation, Stats3.Energy, Stats3.Homogeneity];
T4 = [Stats4.Contrast, Stats4.Correlation, Stats4.Energy, Stats4.Homogeneity];
RUQCN = Stats1.Contrast;
RUQCR = Stats1.Correlation;
RUQEG = Stats1.Energy;
RUQMP = MaxProb1;
RUQEN = Entropy1;
LUQCN = Stats2.Contrast;
LUQCR = Stats2.Correlation;
LUQEG = Stats2.Energy;
LUQEN = Entropy2;
RLQCN = Stats3.Contrast;
RLQCR = Stats3.Correlation;
RLQEN = Entropy3;
LLQCN = Stats4.Contrast;
LLQCR = Stats4.Correlation;

Abnormal_RUQ = ...
	(RUQCN >=18 & RUQCN <= 146) | ...
	(RUQCR >= 0.97 & RUQCR <= 0.99)| ...
	(RUQEG >= 0.31 & RUQEG <=0.51) | ...
	(RUQEN >= 2.91 & RUQEN <=3.79) | ...
	(RUQMP >= 0.57 & RUQMP <= 0.69)
figure, subplot(2,2,1)
imshow(RUQ), title('RUQ')
drawnow;

if Abnormal_RUQ
	xlabel('Abnormal', 'FontSize', 15, 'Color', 'r', 'FontWeight', 'bold')
end

Abnormal_LUQ = ...
	(LUQCN >= 50 & LUQCN <= 78) | ...
	(LUQCR >= 0.97 & LUQCR <= 0.99) | ...
	(LUQEG >= 0.34 & LUQEG <= 0.52) | ...
	(LUQEN >= 2.91 & LUQEN <= 3.79)
subplot(2,2,2), imshow(LUQ), title('LUQ')
if Abnormal_LUQ
	xlabel('Abnormal', 'FontSize', 15, 'Color', 'r', 'FontWeight', 'bold')
end

Abnormal_RLQ = ...
	(RLQCN >= 17 & RLQCN <=157) | ...
	(RLQCR >= 0.98 & RLQCR <= 1.00)
subplot(2,2,3), imshow(RLQ), title('RLQ')
if Abnormal_RLQ
	xlabel('Abnormal', 'FontSize', 15, 'Color', 'r', 'FontWeight', 'bold')
end

Abnormal_LLQ = ...
	(LLQCN >= 21 & LLQCN <= 179) | ...
	(LLQCR >= 0.97 & LLQCR <= 0.99)
subplot(2,2,4), imshow(LLQ), title('LLQ')
if Abnormal_LLQ
	xlabel('Abnormal', 'FontSize', 15, 'Color', 'r', 'FontWeight', 'bold')
end

fprintf('Done running %s.m ...\n', mfilename);

[Nileema Abedeera]

Ohh I was not aware it. Thank you so much!!!

Finally I tried with this changes you have given and got the desired result.

Also I wanted to check if I can give the normal quadrants label "Normal"

so I tried it like this,

if Abnormal_RUQ
	xlabel('Abnormal', 'FontSize', 15, 'Color', 'r', 'FontWeight', 'bold')
end
if ~ Abnormal_RUQ
	xlabel('Normal', 'FontSize', 15, 'Color', 'r', 'FontWeight', 'bold')
end

Is it correct to use it like that?

Thank you again!!!

[Image Analyst]

You can do this:

if Abnormal_RUQ
	xlabel('Abnormal', 'FontSize', 15, 'Color', 'r', 'FontWeight', 'bold')
else
	xlabel('Normal', 'FontSize', 15, 'Color', 'r', 'FontWeight', 'bold')
end

[Nileema Abedeera]

Thank you so much Sir..!

One more thing I need to ask from you,

finally I want these abnormal quadrant image border to be outlined with a red border such as I have shown in the first image on the question.

I used "visboundaries" function just after this xlabel command.

if Abnormal_RUQ
visboundaries (RUQ)
xlabel('Abnormal', 'FontSize', 12, 'Color', 'r', 'FontWeight', 'bold')
end

But it gave something like this.

Would you help me to have an idea how to do it?

I have attached the resulted figure here.

Thank you.

[Image Analyst]

After you call subplot() for the particular one you want to draw on, you need to call hold on so the plot won't blow away the image, then call hold off

subplot(2, 2, 1);
imshow(theImage);
hold on;
plot(x, y, 'r-', 'LineWidth', 2);
hold off;

[Nileema Abedeera]

Yes, it solved my problem!!!

however I couldn't use this line; as I got an error.

Therefore I just used hold on and off with 'visboundaries' and got the roi outlined with red.

plot(x, y, 'r-', 'LineWidth', 2);
error;
Error in Abnormal (line 45)
plot(x, y, 'r-', 'LineWidth', 2);

Thank you so much!!!

[Image Analyst]

You're welcome. Then I guess you can go ahead and "Accept this answer". Thanks in advance.

[Nileema Abedeera]

Yes I've already done it.

Thanks again!!!

[Nileema Abedeera]

Hi again, sorry if I'm bothering you..

I am thinking if I can use my line below;

(RUQCN >=18 & RUQCN <= 146)|... 

in a different way like this;

(RUQCN>=18 : RUQCN<=146)|...

My questions are,

1. Is it a correct way to define the logic here,
2. Will it give the same desired output as the above one?
3. If so, what is the best suitable method of use?

Thank you soo much!

[Image Analyst]

No. Why would you want it to do that? First of all, the <= or >= evaluate to 0 or 1, so that means you'd be constructing a vector going from the first value to the second value in steps of 1. See what happens when you do that:

>> 0:0

ans =

0

>> 0:1

ans =

0 1

>> 1:0

ans =

1×0 empty double row vector

>> 1:1

ans =

1

It just does not make sense to do it.

[Nileema Abedeera]

Yes now I understand.

I was thinking as this 18 and 146 defines the range of values I need to consider and that's why I was worrying if it works.

Thank you so much for the explanation.