how to display images from a nifti file?

I have a liver CT dataset where all images is in .nii file format. Each image file has mulitple slices (say 512*512*500). Using niftiread command i read the image file and now I am visualizing each of the 500 slices one by one using imshow command.

imshow(i = Y(:,:,400)) %400 is my current slice number

Now my image 'i' is class int16 and has values from 0 to 2328. I think these values are Hounsfield units (HU). My image 'i' is like below, where i couldnt see any fine details from the image.

When i referred a paper they said as, 'the Hounsfield unit is applied in window range [100, 400] and then after some pre-processing steps the image is visualized.

how can i do that to bring my image 'i' like a normal CT image?

[Manjunath R V]

Hello emerson can i get a code for hounsfield unit and to enhance the contrast of an ct image

Answer 1

[KALYAN ACHARJYA]

Read the individuals slices and convert them to binary, is there any issue?

slice_test=Y(:,:,400) % Just as example of slice number 400
bw_image=imbinarize(slice_test,method);
%...............................^ any specifice method, 
% or delete method part, Matlab use for default Method

[Emerson Nithiyaraj]

No.if i binarize the slice it just looks like

[KALYAN ACHARJYA]

Yes, the u result images are not binary image, do contrast streching (Gray Level Transformation), do check. I will reply in afternoon session

Answer 2

[Manjunath R V]

even i want clarification like how to apply hounsifield unit to ct image ? for better details

[Emerson Nithiyaraj]

Try this code

a = zeros(size(x));
for i=1:size(a,1)
    for j=1:size(a,2)
        a(i,j) = max(min((x(i,j)), 400),-100 );
    end
end

x(i,j) is the input image.

[Manjunath R V]

thank you for u r response will check the code that u have shared to me

Answer 3

[Manjunath R V]

am working on liver segmentation on CT dataset i need a help regarding labeling of a image since am using sementic segmentation

imDir = fullfile(dataSetDir,'Inp_256'); this is my input original dataset or training data

input image attached here for ref

pxDir = fullfile(dataSetDir,'lab256'); this is my labeled data of the training data

labeled image for the input image attached here

this dataset i found in 3DIRCADB where it contains both training data as well as labeled images and masks for that training data

now my query is to train my network i should use oly training data as well as labelled data rite?

is labelled data whatever they have given is it enough to train my network