Use of Ultrasound and Computer Vision for 3D Reconstruction

The main result of this work is an approach for reconstructing the 3D shape and pose of tumors for applications in laparoscopy from stereo endoscopic ultrasound images using Conformal Geometric Algebra. We record simultaneously stereo endoscopic and ultrasonic images and then the 3D pose of the ultrasound probe is calculated using conformal geometric algebra. When the position in 3D of the ultrasound probe is calculated, we compound multiple 2D ultrasound images into a 3D volume. To segment 2D ultrasound images we have used morphological operators and compared its performance versus the obtained with segmentation using level set methods.