Toward a generic real-time compression correction framework for tracked ultrasound
Intraoperative ultrasound imaging is a commonly used modality for image guided surgery and can be used to monitor changes from pre-operative data in real time. Often a mapping of the liver surface is required to achieve image-tophysical alignment for image guided liver surgery. Laser range scans and tracked optical stylus instruments have both been utilized in the past to create an intraoperative representation of the organ surface. This paper proposes a method to digitize the organ surface utilizing tracked ultrasound and to evaluate a relatively simple correction technique. Surfaces are generated from point clouds obtained from the US transducer face itself during tracked movement. In addition, a surface generated from a laser range scan (LRS) was used as the gold standard for evaluating the accuracy of the US transducer swab surfaces. Two liver phantoms with varying stiffness were tested. The results reflected that the average deformation observed for a 60 second swab of the liver phantom was 3.7 ± 0.9 mm for the more rigid phantom and 4.6 ± 1.2 mm for the less rigid phantom. With respect to tissue targets below the surface, the average error in position due to ultrasound surface digitization was 3.5 ± 0.5 mm and 5.9 ± 0.9 mm for the stiffer and softer phantoms respectively. With the simple correction scheme, the surface error was reduced to 1.1 ± 0.8 mm and 1.7 ± 1.0 mm, respectively; and the subsurface target error was reduced to 2.0 ± 0.9 mm and 4.5 ± 1.8 mm, respectively. These results are encouraging and suggest that the ultrasound probe itself and the acquired images could serve as a comprehensive digitization approach for image guided liver surgery.