lesion_painter: A MATLAB repository from mbolding

A set of Matlab scripts to ease infarct lesion segmentation in 9.4T T2 MRI images (hyperintense) that have poor homogeneity.

main files in this repository

displayTiffImagesMosaic.m: take the TIFF images output from bruker as individual files and make a mosaic for thresholding
findRedPixels.m: finds the area of the segmented lesion. multiply by voxel volume to get the lesion volume approximation
processAllTiffFolders.m: you can use this to run displayTiffImagesMosaic on a folder tree if things are named right
thresholdROI.m: interactive segmentation for poor homogeneity images

instructions for analyzing one volume

1. run displayTiffImagesMosaic.m on the folder of tifs to make a mosaic image of each volume

you can run processAllTiffFolders.m on this step instead to make mosaics of lots of volumes if things are organized correctly. see below.

2. run thresholdROI.m and draw boxes around the lesions on the mosaic volumes to paint them

When your mouse pointer is a cross you can draw an ROI box. When you have a arrow, hit a key (see below) or click the mouse to get a cross. keys:

q: quit

u: undo last ROI, one level deep

d: "decrease", shrink lesion

f: "fill more", grow lesion

e: erase with ROI

s: save the result, calculate areas, and log the result to log.txt

example segmentation

example output: Number of red pixels found: 15583

3. multiply the number of pixels by the voxel volume to get the lesion volume

do this by processing the log file with make_results_table.m

to determine voxel volume

Determine the voxel volume by loading the DICOM into a DICOM viewer (ITKSnap, OsiriX, MRICron, etc.) or into Matlab.

To load DICOM files into MATLAB and find the voxel size, you can follow these steps:

1. Load DICOM Files

MATLAB offers functions to read DICOM files, such as dicomread for reading image data and dicominfo for extracting metadata.

% Specify the directory containing DICOM files
dicomDir = 'path_to_dicom_directory';

% List all files in the directory (assuming they have .dcm extension)
dicomFiles = dir(fullfile(dicomDir, '*.dcm'));

% Load the first DICOM file as an example
filename = fullfile(dicomDir, dicomFiles(1).name);
info = dicominfo(filename);
imageData = dicomread(info);

2. Access Voxel Size Information

Voxel size information is typically stored in the DICOM metadata. Common field names include PixelSpacing and SliceThickness.

% Retrieve voxel size information from the metadata
pixelSpacing = info.PixelSpacing;  % In-plane pixel spacing (row, column) in mm
sliceThickness = info.SliceThickness;  % Thickness of the slices in mm

% Display the voxel size
voxelSize = [pixelSpacing(1), pixelSpacing(2), sliceThickness];
disp(['Voxel size: ', num2str(voxelSize(1)), ' x ', num2str(voxelSize(2)), ' x ', num2str(voxelSize(3)), ' mm']);

% Display the voxel volume
voxelVolume = prod(voxelSize);  % In cubic millimeters
disp(['Voxel volume: ', num2str(voxelVolume), ' cubic mm']);

for our current thinslice T2 protocol...

Voxel size: 0.078125 x 0.078125 x 0.5 mm
Voxel volume: 0.0030518 cubic mm

vox_size.m implements the above.

TIF folder organization

data oraganization for the TIFs is a top level dir, then dir by date, then the sample dirs with the TIFs e.g.:

./trimmed_data/May_24_2024:
M1_TIF			M3_TIF			MRI_Notes.jpeg
M2_TIF			M5_TIF			May_24_2024_Andrabi.zip

./trimmed_data/May_24_2024/M1_TIF:
4_T2_RARE_thinslice_Im01.tif	4_T2_RARE_thinslice_Im07.tif	4_T2_RARE_thinslice_Im13.tif	4_T2_RARE_thinslice_Im19.tif
4_T2_RARE_thinslice_Im02.tif	4_T2_RARE_thinslice_Im08.tif	4_T2_RARE_thinslice_Im14.tif	4_T2_RARE_thinslice_Im20.tif
4_T2_RARE_thinslice_Im03.tif	4_T2_RARE_thinslice_Im09.tif	4_T2_RARE_thinslice_Im15.tif	4_T2_RARE_thinslice_Im21.tif
4_T2_RARE_thinslice_Im04.tif	4_T2_RARE_thinslice_Im10.tif	4_T2_RARE_thinslice_Im16.tif
4_T2_RARE_thinslice_Im05.tif	4_T2_RARE_thinslice_Im11.tif	4_T2_RARE_thinslice_Im17.tif
4_T2_RARE_thinslice_Im06.tif	4_T2_RARE_thinslice_Im12.tif	4_T2_RARE_thinslice_Im18.tif

./trimmed_data/May_24_2024/M2_TIF:
4_T2_RARE_thinslice_Im01.tif	4_T2_RARE_thinslice_Im07.tif	4_T2_RARE_thinslice_Im13.tif	4_T2_RARE_thinslice_Im19.tif...

...

./trimmed_data/May_30_2024:
IMG_1267.JPG		M1_TIF			M3_TIF			May_30_2024_Andrabi.zip

./trimmed_data/May_30_2024/M1_TIF:
4_T2_RARE_thinslice_Im01.tif	4_T2_RARE_thinslice_Im07.tif	4_T2_RARE_thinslice_Im13.tif	4_T2_RARE_thinslice_Im19.tif
4_T2_RARE_thinslice_Im02.tif	4_T2_RARE_thinslice_Im08.tif	4_T2_RARE_thinslice_Im14.tif	4_T2_RARE_thinslice_Im20.tif
4_T2_RARE_thinslice_Im03.tif	4_T2_RARE_thinslice_Im09.tif	4_T2_RARE_thinslice_Im15.tif	4_T2_RARE_thinslice_Im21.tif
4_T2_RARE_thinslice_Im04.tif	4_T2_RARE_thinslice_Im10.tif	4_T2_RARE_thinslice_Im16.tif
4_T2_RARE_thinslice_Im05.tif	4_T2_RARE_thinslice_Im11.tif	4_T2_RARE_thinslice_Im17.tif
4_T2_RARE_thinslice_Im06.tif	4_T2_RARE_thinslice_Im12.tif	4_T2_RARE_thinslice_Im18.tif

./trimmed_data/May_30_2024/M3_TIF:
4_T2_RARE_thinslice_Im01.tif	4_T2_RARE_thinslice_Im07.tif	4_T2_RARE_thinslice_Im13.tif	4_T2_RARE_thinslice_Im19.tif
4_T2_RARE_thinslice_Im02.tif	4_T2_RARE_thinslice_Im08.tif	4_T2_RARE_thinslice_Im14.tif...	

...

mbolding/lesion_painter