Proceedings Article | 3 March 2017
Reza Seifabadi, Sheng Xu, Fereshteh Aalamifar, Peter Pinto, Bradford Wood
KEYWORDS: Prostate, Biopsy, Magnetic resonance imaging, Ultrasonography, Cancer, 3D acquisition, Human-machine interfaces, Robotics, Image fusion, Image registration
Purpose: Fusion of Magnetic Resonance Imaging (MRI) with intraoperative real-time Ultrasound (US) during prostate biopsy has significantly improved the sensitivity of transrectal ultrasound (TRUS) guided cancer detection. Currently, sweeping of the TRUS probe to build a 3D volume as part of the fusion process and the TRUS probe manipulation for needle guidance are both done manually. A motorized, joystick controlled, probe holder was custom fabricated that can potentially reduce inter-operator variability, provide standardization of needle placement, improve repeatability and uniformity of needle placement, which may have impacts upon the learning curve after clinical deployment of this emerging approach. Method: a 2DOF motorized probe holder was designed to provide translation and rotation of a triplane TRUS end firing probe for prostate biopsy. The probe holder was joystick controlled and can assist manipulation of the probe during needle insertion as well as in acquiring a smoother US 2D to 3D sweep in which the 3D US volume for fusion is built. A commercial MRI-US fusion platform was used. Three targets were specified on MR image of a commercial prostate phantom. After performing the registration, two operators performed targeting, once manually and once with the assistance of the motorized probe holder. They repeated these tasks 5 times resulting in a total of 30 targeting events. Time of completion and mechanical error i.e. distance of the target from the needle trajectory in the software user interface were measured. Repeatability in reaching a given target in a systematic and consistent way was measured using a scatter plot showing all targets in the US coordinate system. Pearson product-moment correlation coefficient (PPMCC) was used to demonstrate the probe steadiness during targeting. Results: the completion time was 25±17 sec, 25±24 sec, and 27±15 sec for free hand and 24±10 sec, 22.5±10 sec, and 37±10 sec for motorized insertion, for target 1, 2, and 3, respectively. The mechanical error was 0.75±0.4 mm, 0.45±0.4 mm, and 0.55±0.4 mm, for free hand approach while it was 1.0±0.57 mm, 0.45±0.4 mm, and 0.35±0.25 mm, for motorized approach, for target 1, 2, and 3, respectively. PPMCC remained almost at 1.0 for the motorized approach while having a variation between 0.9 and 1.0 for the free hand approach. Conclusions: motorized fusion guided prostate biopsy in a phantom study was feasible and non-inferior or comparable to the free hand manual approach in terms of accuracy and speed of targeting, while being superior in terms of repeatability and steadiness.
