A simple segmentation method that segments the atrial wall using iterative dilation and intensity thresholding. It requires the intensity distribution information of the ventricular myocardium.
The usage is through command line:
mirtk evaluate-la-wall <input_img> <la_mask> <output_img> <intensity_t1> <intensity_t2> <iterations_n>
Note that the program assumes that the input images are CTA (contrast-enhanced angiography).
This program outputs the la wall segmented from endocardial segmentation based on intensity values [t1,t2] and dilation with n steps.
The program expects a binary mask to be provided as an input parameter <la_mask>, where 1 is set as endocardium. This is the endocardial segmentation of the atrial blood pool.
All input image files are either NifTII or GIPL or HDR Analyze.
the <intensity_t1> and <intensity_t2> are intensity ranges manually selected for the atrial wall. The values t1 and t2 are ideally obtained from ventricular myocardium, +/- 2-standard deviations of the mean of an intensity sample from that region.
The program outputs a binary image with the wall labelled sequentially at each iterative step as 1, 2 and so on.
Dr. Rashed Karim
Department of Biomedical Engineering
King's College London