Source-linked AI summary
Towards Transcervical Ultrasound Image Guidance for Transoral Robotic Surgery
Wanwen Chen, Megha Kalia, Qi Zeng, Emily H. T. Pang, Razeyeh Bagherinasab, Thomas D. Milner, Farahna Sabiq, Eitan Prisman, Septimiu E. Salcudean
TL;DR
TORS is challenging, and the feasibility of transcervical ultrasound guidance and MRI–3D-US registration in the oropharynx remains underexplored. The paper evaluates transcervical 3D-US registration and develops an optically tracked ultrasound-guided AR system. The authors report approximately 27-pixel US-to-camera accuracy and conclude that transcervical 3D US is promising for TORS guidance.
Problem
TORS is challenging without haptic feedback, and the feasibility of transcervical ultrasound and MRI–3D-US registration in the oropharynx remains insufficiently studied.
Source: §1 · pp. 2–3Method
The authors evaluate MRI registration with transcervical 3D US and develop an optically tracked US-to-robot calibration and AR workflow for displaying anatomy in endoscope images.
Source: §1 · p. 3, §2.2 · p. 4Results
The final system's accuracy from US to camera is approximately 27 pixels, and the authors conclude transcervical 3D US is promising for TORS image guidance.
Source: §4 · p. 8, §5 · p. 10Takeaways & Limitations
Transcervical 3D US is a promising approach for image guidance in TORS.
Source: §5 · p. 10MRI–US registration requires manual initialization, and the optical-tracker marker must remain visible to avoid worsening compounded US volume quality.
Source: §4 · p. 9
Abstract
from arXiv · showhide
Purpose: Trans-oral robotic surgery (TORS) using the da Vinci surgical robot is a new minimally-invasive surgery method to treat oropharyngeal tumors, but it is a challenging operation. Augmented reality (AR) based on intra-operative ultrasound (US) has the potential to enhance the visualization of the anatomy and cancerous tumors to provide additional tools for decision-making in surgery. Methods: We propose and carry out preliminary evaluations of a US-guided AR system for TORS, with the transducer placed on the neck for a transcervical view. Firstly, we perform a novel MRI-transcervical 3D US registration study. Secondly, we develop a US-robot calibration method with an optical tracker and an AR system to display the anatomy mesh model in the real-time endoscope images inside the surgeon console. Results: Our AR system reaches a mean projection error of 26.81 and 27.85 pixels for the projection from the US to stereo cameras in a water bath experiment. The average target registration error for MRI to 3D US is 8.90 mm for the 3D US transducer and 5.85 mm for freehand 3D US, and the average distance between the vessel centerlines is 2.32 mm. Conclusion: We demonstrate the first proof-of-concept transcervical US-guided AR system for TORS and the feasibility of trans-cervical 3D US-MRI registration. Our results show that trans-cervical 3D US is a promising technique for TORS image guidance.
1 Introduction
TORS offers a minimally invasive approach to oropharyngeal cancer but is challenging without haptic feedback. The introduction motivates transcervical ultrasound guidance and proposes an MRI–3D-US registration study and proof-of-concept AR system.
1 Introduction
- Image guidance could improve visualization of anatomy and tumors to support resection-margin decisions and reduce postoperative hemorrhage risk.TORS requires surgeons to work through the mouth without haptic feedback and with detailed knowledge of oropharyngeal anatomy.Source: §1 · p. 2
- Current TORS guidance mostly uses preoperative CT or MRI, while intraoperative CT can improve fiducial localization and task efficiency but adds radiation risk and occupies operating-room space.Preoperative CT may not represent deformed anatomy sufficiently.Source: §1 · p. 2
- Oral-cavity ultrasound can require removing TORS tools or using a miniaturized transducer with limited scanning range; transcervical ultrasound permits continuous imaging, but its feasibility in TORS remains unstudied.Transcervical 3D ultrasound has been used in neck procedures and oral cancer diagnosis.Source: §1 · p. 2
- MRI offers soft-tissue contrast and cancer imaging, motivating MRI–ultrasound registration; this work evaluates oropharyngeal MRI-to-freehand-3D-US registration, preoperative-to-intraoperative US registration, and transcervical US-guided TORS AR.The feasibility of MRI–3D-US registration in the oropharynx is not well studied.Source: §1 · pp. 2–3
2 Materials and Methods
The study evaluates an ultrasound-guided AR workflow for TORS, combining MRI–ultrasound registration with system calibration and preoperative-to-intraoperative ultrasound registration. It uses tracked ultrasound and robot hardware to display an MRI-derived anatomy model in endoscopic images.
2 Materials and Methods
- The AR workflow combines preoperative MRI and US planning, initial anatomy display after MRI–US registration, and intraoperative US–US registration to update the model.During surgery, the patient’s tongue is retracted and the surgeon scans the tumor-side neck with the US transducer.Source: §2 · p. 3
2.1 System Hardware
- The system uses a da Vinci S robot, a BK3500 with a 14L3 linear transducer, and an optical tracker; a printed holder lets the robot’s third arm maneuver the probe.The holder is grasped with the da Vinci ProGrasp.Source: §2.1 · p. 3, Figure 3 · p. 5
2.2 System Calibration
- System calibration composes transformations from the US image frame through the probe marker and robot frame to the endoscopic camera image.The camera projection matrix includes the hand-eye calibration matrix and camera intrinsic matrix.Source: Figure 2 · p. 4, §2.2 · p. 4
- US-to-tracker calibration uses synchronized tracked US images, a tracked stylus, and manually selected stylus-tip locations to estimate the probe-to-US transformation.A customized marker and corresponding tracker and robot points are used to estimate the robot-to-tracker rigid transformation.Source: §2.2 · p. 4
2.3 MRI-3D US Registration
- MRI–US registration assumes an affine transformation, with manual rigid preregistration refined using LC2 optimization.The study evaluates a 3D transducer in five healthy volunteers and four patients, and freehand 3D US in three healthy volunteers.Source: §2.3 · p. 5
2.4 Preoperative US-Intraoperative US Registration
- Preoperative-to-intraoperative registration uses freehand 3D US acquired from two TORS patients before and after tongue suspension.QuickSyN ANTs minimizes mean square error after a translation initialization.Source: §2.4 · p. 5
3 Experiments and Results
The evaluations measure US-to-camera calibration, MRI-to-US registration, and preoperative-to-intraoperative US registration. Reported passages specify calibration errors and evaluation procedures, while the registration tables and figures are identified without their numerical results.
3.1 System Evaluation
- 1.84 ± 0.57 mm was the testing mean projection error for robot-tracker calibration, while stylus-based calibration testing error was 0.78 ± 0.41 mm.Robot-tracker calibration used 100 paired points split evenly between estimation and testing; stylus-based calibration used 25 points for estimation and 17 for testing.Source: §3.1 · pp. 5–6
3.2 Feasibility Evaluation of MRI-US Registration
- MRI-to-3D-US evaluation measured target registration error for probe-collected US and both target registration error and vessel-centerline distance for freehand US.The freehand scan covered the neck from the common carotid to the submandibular gland; the supplied passages refer to Tables 2 and 3 but do not state their numerical results.Source: §3.2 · p. 7, Table 2 · p. 7, Table 3 · p. 7
3.3 Feasibility Evaluation of US-US Registration
- Preoperative-to-intraoperative 3D-US evaluation reports TRE, vessel-centerline distance, maximum deformation, and runtime, but the supplied passages do not give their values.Figure 8 shows sagittal and axial views before and after registration; the freehand acquisition used 9 MHz imaging at 4 cm depth with reconstructed volume resolution of 0.2 mm3.Source: §3.3 · pp. 7–8, Table 4 · p. 8, Figure 8 · p. 8
4 Discussion
The discussion reports system calibration accuracy and registration findings, including deformation from tongue suspension. It also identifies limitations in registration initialization, tracking, data volume, runtime, and probe visibility.
4 Discussion
- Approximately 27 pixels (2.5% of the image diagonal) was the US-to-camera system accuracy; overall system error was 9.93 mm, comparable with the 7.6 mm error reported for preoperative CT-to-tracker registration.The da Vinci S camera calibration contributes substantially to the error; the authors expect this to be reduced with the pre-calibrated camera of the da Vinci Xi system.Source: §4 · pp. 8–9
- Tongue suspension can cause 10–20 mm deformations; intraoperative US-US registration accommodated them with maximum deformation of 10–15 mm and deformable-registration error of 4–12 mm.The final system error will include both system calibration error and deformable-registration error, which the authors say needs future evaluation.Source: §4 · p. 9
- MRI-US registration requires manual initialization, optical-marker visibility, and reliable tracking; US-US evaluation used little data and its deformable registration cannot run near real-time.Surgeons cannot see the US probe, although an assistant can maneuver it; the authors propose an external camera view in the console.Source: §4 · p. 9
5 Conclusion
The work presents a first proof-of-concept transcervical US-guided AR system for TORS and evaluates transcervical 3D US-MRI and preoperative-intraoperative US registration. It concludes that transcervical 3D US is promising for TORS image guidance.
5 Conclusion
- The work presents the first proof-of-concept transcervical US-guided AR system for TORS.Source: §5 · p. 9
- The feasibility study evaluates transcervical 3D US-MRI registration using two types of 3D US and preoperative-intraoperative US registration.Source: §5 · pp. 9–10
- Transcervical 3D US is a promising approach for TORS image guidance, with potential to visualize anatomy and correct distortion intraoperatively.Source: §5 · p. 10
Declarations
The declarations identify the study's funding, competing interests, ethics approval, and informed consent.
Declarations
- The study was funded by an NSERC Discovery Grant and the Charles Laszlo Chair in Biomedical Engineering held by Dr. Salcudean.Source: Declarations · p. 10
- The authors report no conflicts of interest.Source: Declarations · p. 10
- The study received institutional ethics approval (#H19-04025), and informed consent was obtained from all participants.Source: Declarations · p. 10