Operations: users would need to operate below commands in MATLAB after adding the scripts to search path.
- Input a data file by pressing 'i';
- Scroll wheel to select a transverse slice where tumor lies.
- Select tumor zone using left button to draw a green line and background zone using right button to draw a read line. The thickness of lines can be adjusted using '+'/'-';
- Select a searching zone of tumor/ROI using 'f' to reduce calculation. 'f' can be use several times;
- Type 'r' to run segmentation;
- Type 'c' to clean the mask if the result is not good;
- When tumor ROIs on all slices have been segmented, press 'o' to output masks to the data folder.
h: help window;
i: open a new nifti file (*.nii, *.img, *.nii.gz);
left mouse draw: mark foreground with green brush;
right mouse draw: mark background with red brush;
f: specify a mask of ROI to reduce computation;
c: clean label and masks of the current slice;
a, d, leftarrow, rightarrow: adjust brush size;
w, s, uparrow, downarrow, wheelup, wheeldown: change slice;
+, -: adjust fill hole size;
t: switch among different semi-supervised learning methods;
r: run semi-supervised learning to segment tumor on the current slice;
m: show/hide masks;
o: output to a mask file;
Hint: select checkboxes to change view along with the main image
To make scrolling more smooth, select only the wanted plots.