HX40? MICRONTRACKER? 雙目相機(jī)

Target registration error reduction for percutaneous abdominal intervention

Author links open overlay panelMateusz?Bas,?Krzysztof?Król,?Dominik?Spinczyk

Show more

Add to Mendeley

Share

Cite

https://doi.org/10.1016/j.compmedimag.2020.101839Get rights and content

Abstract

A real-time methodology that finds spatio-temporal correspondence between the positions of the target point in the pre-treatment 3DCT image and during the procedure was proposed. It based on minimizing the target registration error in III tier registration circuits. Particle Swarm Optimization and Differential Evaluation were used to find optimal values of Elastic Body Spline parameters in the generation of abdominal deformation field. Different transformation classes have been tested: rigid, affine, Thin Plate Spline, Elastic Body Spline. The lowest TRE was obtained for the swarm optimization algorithm - differential evolution for the rigid and affine version: 3.47 and 3.73 mm, respectively.

Graphical abstract

1. Download :?Download high-res image (248KB)

2. Download :?Download full-size image

Introduction

Estimation of target position during minimally invasive abdominal procedures is one of the main challenges of modern developing imaging navigation systems. Image-based navigation systems utilize knowledge about the patient's anatomy, acquired via diagnostic images, during therapy. Given the current state-of-the-art, the location of the pathological lesion can be effectively estimated for rigid structures. However, it is a challenge for parenchymal structures (Kenngott et al., 2014; Neshat et al., 2013; Phee and Yang, 2010; Spinczyk, 2015).

Knowing the information about the positions of the fiducial markers placed on the patients skin and visible in the preoperative imaging marker-based registration can be performed to match the reference frames of preoperative volumetric imaging with the position of the patient during surgery (Fabian and Spinczyk, 2018). Unfortunately, Fitzpatrick found no statistical correlation between Fiducial Registration Error (FRE) and Target Registration Error (TRE) (Fitzpatrick, 2009) and proved that using only one chain of registration may lead to underestimation of registration errors (Fitzpatrick, 2001). This is why the Assessing Quality Using Image Registration Circuits method (AQUIRC), proposed by Datteri (R. D. Datteri and Dawant, 2012a, 2012b), used III order registration circuits to register preoperative and simulated intraoperative data, by making modifications to the positions of the markers with estimated Fiducial Localization Error (FLE), and then making the registration between the modified markers positions. This has since been used by Datteri & Dawant (Datteri and Dawant, 2012a, 2012b) to estimate the quality of fit for rigid image registration.

In modeling of deformations, the mechanical modeling takes into account the distribution of forces acting on internal organs. But during a procedure the direct values of the forces acting on the tissues are not known. The commonly known finite element method (FEM) is not used, but the simulation is based on Position Based Simulation (PBS), where it is not necessary to specify explicit values of the forces (Camara et al., 2016). Position based simulations used in a minimally invasive surgery scenario provide a different approach for registration of the preoperative imaging with the laparoscopic imaging. Taking into account the modelling of the gravity, the pressure of the gas in the abdominopelvic cavity along with visual cues in form crucial anatomical landmarks the registration is performed in closer range of the target lesion inside an organ thus providing more reliable positioning of the preoperative imaging (?zgür et al., 2018). However, repeatability of such an approach is difficult due to different conditions and visibility of the laparoscopic surgical scene. In this work, the discussed methods were not necessary, because the repeatability of the patient's position before and during the procedure was ensured. Details can be found in Section 2.4.

The aim of the work is to propose a method that finds spatio-temporal correspondence between the positions of the target point in the pre-treatment 3D Computed Tomography (CT) image and during the procedure. Considering the largest source of deformations in the abdominal cavity is the respiratory process, a method of selecting the respiratory phase was proposed, the most suitable to insert the tool. This selection was based on minimizing the TRE error in III-tier registration circuits. Then, the use of Particle Swarm Optimization and Differential Evaluation methods to find optimal values of Elastic Body Spline parameters in the generation of abdominal deformation field was presented.

The practical goal of the presented approach is the clinical verification of the possibility of using abdominal surface markers’ position to estimate target position during percutaneous abdominal intervention.

Section snippets

Materials and methods

The stages of proposed methodology are presented in Fig. 1. The first stage is to choose the respiratory phase that best corresponds to preoperative CT (Fig. 1a)(point 2.1 Selection of respiratory phase). The second stage is the construction of the abdominal deformation field model based on Elastic Body Splines (Fig. 1b) (point 2.2 Estimation of target position during treatment). The third step is to find the value of the Elastic Body Spline curve parameters using the Particle Swarm

Results

Table 2 summarizes the obtained numerical values of the various classes of transformation. When training methods on the respiratory phase found by a method based on registration circuits, the lowest TRE was obtained for the Swarm Optimization algorithm - Differential Evolution (DE) for both the rigid version – Table 2: 3.47 mm1)?(RIGID-EBSDE-AQUIRC), 3.73 mm2)?(RIGID-EBSPSO-AQUIRC) and affine 4.412 mm3)?(AFFINE-EBSDE-AQUIRC), 4.697 mm4)?(AFFINE-EBSPSO-AQUIRC). Acquired results are also

Discussion

Due to the lack of information about the actual location of the target during the procedure, image navigation systems try to use information about the spatial location of markers and the target from preoperative layered data. In addition, this approach uses the assumption of periodic deformities associated with respiratory movement, which is supported by the fact that the patient breathes regularly using a respirator. With this approach, the problem of estimating the location of the target

Conclusions

The real-time method that finds spatio-temporal correspondence between the positions of the target point in the pre-treatment 3DCT image and during the procedure was proposed. The results obtained indicate the possibility of clinical application, including in the future verification of the accuracy of the method, where the location of the target point will be inside the abdominal organ.

Funding sources

This research was supported by?Silesian University of Technology statutory?financial support No.?BKM-671/RIB1/2020?(07/010/BKM20/0047).

CRediT authorship contribution statement

Mateusz Bas:?Investigation, Writing - original draft, Visualization.?Krzysztof Król:?Investigation, Software, Validation, Formal analysis.?Dominik Spinczyk:?Investigation, Writing - review & editing, Methodology.

Declaration of Competing Interest

There is no conflict of interest.

Acknowledgments

The method was evaluated based on imaging data from patients treated at the Second Department of Clinical Radiology and the Department of General, Transplant and Liver Surgery at the Medical University of Warsaw after acceptance by the ethics committee of the Medical University of Warsaw (KB/33/2018). Principles stated in Declaration of Helsinki on Ethical Principles for Medical Research Involving Human Subjects were followed.