Value-assessment of computer-assisted navigation strategies during percutaneous needle placement

Abstract

Purpose: Navigational strategies create a scenario whereby percutaneous needle-based interventions of the liver can be guided using both pre-interventional 3D imaging datasets and dynamic interventional ultrasound (US). To score how such technologies impact the needle placement process, we performed kinematic analysis on different user groups.

Methods: Using a custom biopsy phantom, three consecutive exercises were performed by both novices and experts (n = 26). The exercise came in three options: (1) US-guidance, (2) US-guidance with pre-interventional image-registration (US + Reg) and (3) US-guidance with pre-interventional image-registration and needle-navigation (US + Reg + Nav). The traveled paths of the needle were digitized in 3D. Using custom software algorithms, kinematic metrics were extracted and related to dexterity, decision making indices to obtain overall performance scores (PS).

Results: Kinematic analysis helped quantifying the visual assessment of the needle trajectories. Compared to US-guidance, novices yielded most improvements using Reg (PS_avg(US) = 0.43 vs. PS_avg(US+Reg) = 0.57 vs. PS_{avg(US+Reg+Nav)} = 0.51). Interestingly, the expert group yielded a reversed trend (PS_avg(US) = 0.71 vs PS_avg(US+Reg) = 0.58 vs PS_{avg(US+Reg+Nav)} = 0.59).

Conclusion: Digitizing the movement trajectory allowed us to objectively assess the impact of needle-navigation strategies on percutaneous procedures. In particular, our findings suggest that these advanced technologies have a positive impact on the kinematics derived performance of novices.

Keywords: Computer-assisted surgery; Image fusion; Navigation; Needle guidance; Performance assessment.

Fig. 1

a A schematic view of the phantom based experimental set-up used to study the impact of the additional computer-assisted navigation technology during an image-guided biopsy. (1) The ultrasound system, (2) EM field generator, (3) phantom including imitation lesions, (4) EM active tracker and fiducial tracker, (5) biopsy needle with fiducials, (6) US probe including fiducials and (7) optical near infrared camera. b US display including navigational strategies

Fig. 2

Example image acquired from the customized abdominal phantom by using a PET-CT with Hounsfield units of: 368 HU for bone, 138 HU for the lesions and − 166 for the ballistic gel, and b ultrasonography

Fig. 3

Tracked needle tip paths depicted in a 3D graph 9of all three assignments, US guided (a), US + Reg guided (b) and US + Reg + Nav guided (c) of both an expert (i) and a novice (ii). The color bar indicates the movement speed at each point within the path

Fig. 4

Feature correlations: a Abrupt speed (i) and acceleration (ii) changes of the needle path are linearly correlated with retractions with R² values of 0.83 and 0.88, respectively. The color bar indicates the density occurrence of corrections and retractions. b Total feature correlation using sPLS-DA analysis

Fig. 5

a A comparative overview of the dexterity and fluency features of the needle for the additional technology (US guided biopsy (red), US + Reg guided biopsy (blue) and US + Reg + Nav guided (green)) for both experts (gray) and novices (white), where *indicates a significance of p < 0.05 between two exercises. b The Dx index given to both experts and novices for each of these technologies

Fig. 6

a An overview of the general features and handling errors to compare the additional technology (US guided biopsy (red), US + Reg guided biopsy (blue) and US + Reg + Nav guided (green)) for both experts (gray) and novices (white), where the significance is indicates by *; p < 0.05, **; p < 0.01 and ***; p < 0.001. b The DM index given to both experts and novices for each of these technologies

Fig. 7

The performance score of each of the participant (gray: experts and white: novices) for each of the technologies; US, US + Reg and US + Reg + Nav. The red line indicates the proficiency level equal to 0.68