METHOD FOR AUTOMATIC SEGMENTATION OF CORONARY SINUS

Method (20), executed by a computer (5), for identifying a coronary sinus of a patient (11), comprising: receiving (S01) a 3D image (15) of a body region of the patient (11); extracting (S03) 2D axial images (13a) of the 3D image (15) taken along respective axial planes, 2D sagittal images (13b) of the 3D image (15) taken along respective sagittal planes, and 2D coronal images (13c) of the 3D image (15) taken along respective coronal planes; applying (S05) an axial neural network (22a) to each 2D axial image (13a) to generate a respective 2D axial probability map (24a), a sagittal neural network (22b) to each 2D sagittal image (13b) to generate a respective 2D sagittal probability map (24b), and a coronal neural network (22c) to each 2D coronal image (13c) to generate a respective 2D coronal probability map (24c); generating (S07), based on the 2D probability maps (24a, 24b, 24c), a 3D mask (30) of the coronary sinus of the patient (11) .