Robot-assisted Catheterization Research Articles

Endovascular Tool Segmentation with Multi-lateral Branched Network during Robot-assisted Catheterization.

Robot-assisted catheterization is routinely carried out for intervention of cardiovascular diseases. Meanwhile, the success of endovascular tool navigation depends on visualization and tracking cues available in the robotic platform. Currently, real-time motion analytics are lacking, while poor illumination during fluoroscopy affects existing physics- and learning-based methods used for tool segmentation. A multi-lateral branched network (MLB-Net) is herein proposed for tool segmentation in cardiovascular angiograms. The model has an encoder with multi-lateral separable convolutions and a pyramid decoder. Model training and validation are done on 1320 angiograms obtained during robot-assisted catheterization in rabbit. Model performance, explained with F1-score of 89.01% and mean intersection-over-union of 90.05% on 330 frames, indicates the model's robustness for guidewire segmentation in angiograms. The MLB-Net offers better performance than the state-of-the-art segmentation models such as U-Net, U-Net++ and DeepLabV3. Thus, it could provide basis for endovascular tool tracking and surgical scene analytics during cardiovascular interventions.

Weighting-Based Deep Ensemble Learning for Recognition of Interventionalists’ Hand Motions During Robot-Assisted Intravascular Catheterization

Robot-assisted intravascular interventions have evolved as unique treatments approach for cardiovascular diseases. However, the technology currently has low potentials for catheterization skill evaluation, slow learning curve, and inability to transfer experience gained from manual interventions. This study proposes a new weighting-based deep ensemble model for recognizing interventionalists' hand motions in manual and robotic intravascular catheterization. The model has a module of neural layers for extracting features in electromyography data, and an ensemble of machine learning methods for classifying interventionalists' hand gestures as one of the six hand motions used during catheterization. A soft-weighting technique is applied to guide the contributions of each base learners. The model is validated with electromyography data recorded during <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">in-vitro</i> and <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">in-vivo</i> trials and labeled as <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">many-to-one</i> sequences. Results obtained show the proposed model could achieve 97.52% and 47.80% recognition performances on test samples in the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">in-vitro</i> and <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">in-vivo</i> data, respectively. For the latter, transfer learning was applied to update weights from the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">in-vitro</i> data, and the retrained model was used for recognizing the hand motions in the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">in-vivo</i> data. The weighting-based ensemble was evaluated against the base learners and the results obtained shows it has a more stable performance across the six hand motion classes. Also, the proposed model was compared with four existing methods used for hand motion recognition in intravascular catheterization. The results obtained show our model has the best recognition performances for both the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">in-vitro</i> and <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">in-vivo</i> catheterization datasets. This study is developed toward increasing interventionalists' skills in robot-assisted catheterization.

Exploring Operators’ Natural Behaviors to Predict Catheterization Trial Outcomes in Robot-Assisted Intravascular Interventions

Recently, robotic catheterization has enhanced the outcomes of cardiovascular interventions. Meanwhile, the roles of operator’s natural behavior in the robot-assisted intravascular procedures need more attention. In this paper, operators’ hand activities related to endovascular tool manipulation are studied to explore how operators’ hand motions aid robotic catheterization. Controlled in-vivo studies were set up to acquire four types of operators’ natural behaviors during 60 robotic catheterization trials, and activity signals were recorded to proxy operators’ skills. A multi-layer recognition model is developed for recognizing the hand motions made during the procedures. The model operates convolution and dense layers to multiplex features extracted in single to multiple data modalities. Starting with initial-decision layer, the model is built to train and classify catheterization trials recorded from nine interventionists as successful and unsuccessful. Next, a motion-decision layer is built to recognize interventionists’ hand motions using features from different data modalities. Lastly, a mixed-decision layer is integrated to recognize the motion patterns of the successful and unsuccessful trials. Results show that the initial-decision layer has an accuracy of 99.44% in predicting the catheterization trials as successful and unsuccessful, while the motion-decision layer shows accuracies of 98.55% and 98.44% in classifying the seven types of hand motions that operators engage during successful and unsuccessful trials, respectively. Also, the mixed-decision layer has an accuracy of 93.96% in recognizing 14 mixed patterns from both trial classes. This study provides an objective template for skill training and evaluation in robot-assisted catheterization.

Study on real-time force feedback for a master-slave interventional surgical robotic system.

In robot-assisted catheterization, haptic feedback is important, but is currently lacking. In addition, conventional interventional surgical robotic systems typically employ a master-slave architecture with an open-loop force feedback, which results in inaccurate control. We develop herein a novel real-time master-slave (RTMS) interventional surgical robotic system with a closed-loop force feedback that allows a surgeon to sense the true force during remote operation, provide adequate haptic feedback, and improve control accuracy in robot-assisted catheterization. As part of this system, we also design a unique master control handle that measures the true force felt by a surgeon, providing the basis for the closed-loop control of the entire system. We use theoretical and empirical methods to demonstrate that the proposed RTMS system provides a surgeon (using the master control handle) with a more accurate and realistic force sensation, which subsequently improves the precision of the master-slave manipulation. The experimental results show a substantial increase in the control accuracy of the force feedback and an increase in operational efficiency during surgery.

Design and characteristics evaluation of a novel VR-based robot-assisted catheterization training system with force feedback for vascular interventional surgery

Vascular interventional surgery has been become a specialized surgical technique because it can minimize the trauma and shorten the recovery time, but it needs more accurate operation and image to guide the surgery. In the meantime, it needs that the doctor has a great deal of experience. Moreover, it needs to train the novice because of the lack of experienced neurosurgeons. In order to solve aforementioned problems, the training system for vascular interventional surgery was proposed. A novel VR-based robot-assisted catheterization training system is designed to solve these problems. This system is composed of the virtual environment and the haptic device. The virtual environment included a catheter model and a vascular model to realize the synchronous movement with the haptic device. In order to imitate soft tissue deformation, mass-spring model has been proposed in physical modeling. In this paper we established a dynamics equation of the moving catheter. We analyzed the elasticity distribution of the wall of vascular according to the previous research. The haptic interaction device based on magneto-rheological fluid as the master manipulator was applied to control the movement of the catheter model in the virtual environment and realize haptic feedback. At the same time, a force feedback calculation model for the novel master manipulator has been proposed. Ten times experiments both in axial movement direction and in radial movement respectively, are carried out, there was within 0.74 mm error in axial movement direction and 3.5° error in radial movement direction. When the moving speed of the catheter is 5 mm/s, the force error between the virtual environment and the feedback force of the damper is within 0.3 mN. The error is in the range which is permitted during VIS. The proposed method improved accurate operability and traceability in this novel VR robotic catheter training system. It offers users better visualization and control. The experimental results indicate that the performance of the synchronous movement and the force feedback can meet our design requirement. The developed training system can be used to train novice for VIS.

Abstract No. 353 - Evaluation of robot-assisted catheterization using the Magellan system for embolization

No. 353 Evaluation of robot-assisted catheterization using the Magellan system for embolization G. Rao, S. Rao; Medicine, University of North Texas Health Science Center, Fort Worth, TX; SierraProvidence Health Network, El Paso, TX Purpose: To evaluate the safety and efficacy of robot-assisted catheter guidance by the Magellan (Hansen Medical, Mountain View, California, USA) robotic catheter for embolization procedures. The following four procedures were evaluated: uterine fibroid embolization, transarterial chemoembolization with drug-eluting beads for hepatocellular carcinoma, embolization for gastric hemorrhage and embolization of pulmonary arteriovenous malformation. Materials and Methods: Four patients underwent embolization using the remotely controlled Magellan catheter system. Documented performance metrics and data include: fluoroscopy time, success of target vessel cannulation, procedural time, contrast use, and overall technical success. Robotic manipulation routes included femoral and radial arterial access and femoral venous access. Assessment included evaluation of embolization ability using a variety of embolic materials. Results: Technical success was 100%. No complications seen. For the four cases: 1) UFE: 6Fr robotic catheter successfully accessed contralateral anterior branch of the internal iliac and ostium of the ipsilateral internal iliac. Further distal cannulation achieved with microcatheter. 2) 6Fr catheter for DEBTACE demonstrated successful catheterization of the right hepatic artery. Distal branches required microcatheter placement. 3) 6Fr catheter via radial access demonstrated successful catheterization of the gastroepiploic artery for particle and coil embolization for gastric bleed. Left gastric artery was poorly visualized and not accessed. 4) 9Fr catheter usage for embolization of pulmonary arteriovenous malformation demonstrated adequate access into right lower lobe pulmonary branch. Distal catheterization of feeding vessel required microcatheter for coil embolization. Reduction in radiation exposure to the operator was over 70% compared to control radiation monitor at patient bedside. Conclusion: Technical success without complication can be achieved during embolization procedures with the robotic system. Distal subselective target embolization may require the use of manual cannulation using a separate microcatheter. Usage of the remotely controlled catheter reduces radiation exposure to the operator.